src/Network/Probe.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Network;
  23
  24 use DOMDocument;
  25 use DomXPath;
  26 use Friendica\Core\Cache\Duration;
  27 use Friendica\Core\Logger;
  28 use Friendica\Core\Protocol;
  29 use Friendica\Database\DBA;
  30 use Friendica\DI;
  31 use Friendica\Model\Contact;
  32 use Friendica\Model\Profile;
  33 use Friendica\Protocol\ActivityNamespace;
  34 use Friendica\Protocol\ActivityPub;
  35 use Friendica\Protocol\Email;
  36 use Friendica\Protocol\Feed;
  37 use Friendica\Util\Crypto;
  38 use Friendica\Util\Network;
  39 use Friendica\Util\Strings;
  40 use Friendica\Util\XML;
  41
  42 /**
  43  * This class contain functions for probing URL
  44  */
  45 class Probe
  46 {
  47         private static $baseurl;
  48         private static $istimeout;
  49
  50         /**
  51          * Rearrange the array so that it always has the same order
  52          *
  53          * @param array $data Unordered data
  54          *
  55          * @return array Ordered data
  56          */
  57         private static function rearrangeData($data)
  58         {
  59                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
  60                                 "community", "keywords", "location", "about", "gender", "hide",
  61                                 "batch", "notify", "poll", "request", "confirm", "poco",
  62                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  63                                 "priority", "network", "pubkey", "baseurl"];
  64
  65                 $newdata = [];
  66                 foreach ($fields as $field) {
  67                         if (isset($data[$field])) {
  68                                 $newdata[$field] = $data[$field];
  69                         } else {
  70                                 $newdata[$field] = "";
  71                         }
  72                 }
  73
  74                 // We don't use the "priority" field anymore and replace it with a dummy.
  75                 $newdata["priority"] = 0;
  76
  77                 return $newdata;
  78         }
  79
  80         /**
  81          * Check if the hostname belongs to the own server
  82          *
  83          * @param string $host The hostname that is to be checked
  84          *
  85          * @return bool Does the testes hostname belongs to the own server?
  86          */
  87         private static function ownHost($host)
  88         {
  89                 $own_host = DI::baseUrl()->getHostname();
  90
  91                 $parts = parse_url($host);
  92
  93                 if (!isset($parts['scheme'])) {
  94                         $parts = parse_url('http://'.$host);
  95                 }
  96
  97                 if (!isset($parts['host'])) {
  98                         return false;
  99                 }
 100                 return $parts['host'] == $own_host;
 101         }
 102
 103         /**
 104          * Probes for webfinger path via "host-meta"
 105          *
 106          * We have to check if the servers in the future still will offer this.
 107          * It seems as if it was dropped from the standard.
 108          *
 109          * @param string $host The host part of an url
 110          *
 111          * @return array with template and type of the webfinger template for JSON or XML
 112          * @throws HTTPException\InternalServerErrorException
 113          */
 114         private static function hostMeta($host)
 115         {
 116                 // Reset the static variable
 117                 self::$baseurl = '';
 118
 119                 $ssl_url = "https://".$host."/.well-known/host-meta";
 120                 $url = "http://".$host."/.well-known/host-meta";
 121
 122                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 123
 124                 Logger::log("Probing for ".$host, Logger::DEBUG);
 125                 $xrd = null;
 126
 127                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 128                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 129                 if ($curlResult->isSuccess()) {
 130                         $xml = $curlResult->getBody();
 131                         $xrd = XML::parseString($xml, false);
 132                         $host_url = 'https://'.$host;
 133                 } elseif ($curlResult->isTimeout()) {
 134                         Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG);
 135                         self::$istimeout = true;
 136                         return false;
 137                 }
 138
 139                 if (!is_object($xrd)) {
 140                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 141                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 142                         if ($curlResult->isTimeout()) {
 143                                 Logger::info('Probing timeout', ['url' => $url], Logger::DEBUG);
 144                                 self::$istimeout = true;
 145                                 return false;
 146                         } elseif ($connection_error && $ssl_connection_error) {
 147                                 self::$istimeout = true;
 148                                 return false;
 149                         }
 150
 151                         $xml = $curlResult->getBody();
 152                         $xrd = XML::parseString($xml, false);
 153                         $host_url = 'http://'.$host;
 154                 }
 155                 if (!is_object($xrd)) {
 156                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
 157                         return [];
 158                 }
 159
 160                 $links = XML::elementToArray($xrd);
 161                 if (!isset($links["xrd"]["link"])) {
 162                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
 163                         return [];
 164                 }
 165
 166                 $lrdd = ['application/jrd+json' => $host_url . '/.well-known/webfinger?resource={uri}'];
 167
 168                 foreach ($links["xrd"]["link"] as $value => $link) {
 169                         if (!empty($link["@attributes"])) {
 170                                 $attributes = $link["@attributes"];
 171                         } elseif ($value == "@attributes") {
 172                                 $attributes = $link;
 173                         } else {
 174                                 continue;
 175                         }
 176
 177                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 178                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 179
 180                                 $lrdd[$type] = $attributes["template"];
 181                         }
 182                 }
 183
 184                 self::$baseurl = $host_url;
 185
 186                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
 187
 188                 return $lrdd;
 189         }
 190
 191         /**
 192          * Perform Webfinger lookup and return DFRN data
 193          *
 194          * Given an email style address, perform webfinger lookup and
 195          * return the resulting DFRN profile URL, or if no DFRN profile URL
 196          * is located, returns an OStatus subscription template (prefixed
 197          * with the string 'stat:' to identify it as on OStatus template).
 198          * If this isn't an email style address just return $webbie.
 199          * Return an empty string if email-style addresses but webfinger fails,
 200          * or if the resultant personal XRD doesn't contain a supported
 201          * subscription/friend-request attribute.
 202          *
 203          * amended 7/9/2011 to return an hcard which could save potentially loading
 204          * a lengthy content page to scrape dfrn attributes
 205          *
 206          * @param string $webbie    Address that should be probed
 207          * @param string $hcard_url Link to the hcard - is returned by reference
 208          *
 209          * @return string profile link
 210          * @throws HTTPException\InternalServerErrorException
 211          */
 212         public static function webfingerDfrn($webbie, &$hcard_url)
 213         {
 214                 $profile_link = '';
 215
 216                 $links = self::lrdd($webbie);
 217                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
 218                 if (!empty($links) && is_array($links)) {
 219                         foreach ($links as $link) {
 220                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
 221                                         $profile_link = $link['@attributes']['href'];
 222                                 }
 223                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
 224                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 225                                 }
 226                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 227                                         $hcard_url = $link['@attributes']['href'];
 228                                 }
 229                         }
 230                 }
 231                 return $profile_link;
 232         }
 233
 234         /**
 235          * Check an URI for LRDD data
 236          *
 237          * this is a replacement for the "lrdd" function.
 238          * It isn't used in this class and has some redundancies in the code.
 239          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 240          *
 241          * @param string $uri Address that should be probed
 242          *
 243          * @return array uri data
 244          * @throws HTTPException\InternalServerErrorException
 245          */
 246         public static function lrdd($uri)
 247         {
 248                 $lrdd = self::hostMeta($uri);
 249                 $webfinger = null;
 250
 251                 if (is_bool($lrdd)) {
 252                         return [];
 253                 }
 254
 255                 if (!$lrdd) {
 256                         $parts = @parse_url($uri);
 257                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 258                                 return [];
 259                         }
 260
 261                         $host = $parts["host"];
 262                         if (!empty($parts["port"])) {
 263                                 $host .= ':'.$parts["port"];
 264                         }
 265
 266                         $path_parts = explode("/", trim($parts["path"], "/"));
 267
 268                         $nick = array_pop($path_parts);
 269
 270                         do {
 271                                 $lrdd = self::hostMeta($host);
 272                                 $host .= "/".array_shift($path_parts);
 273                         } while (!$lrdd && (sizeof($path_parts) > 0));
 274                 }
 275
 276                 if (!$lrdd) {
 277                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
 278                         return [];
 279                 }
 280
 281                 foreach ($lrdd as $type => $template) {
 282                         if ($webfinger) {
 283                                 continue;
 284                         }
 285
 286                         $path = str_replace('{uri}', urlencode($uri), $template);
 287                         $webfinger = self::webfinger($path, $type);
 288
 289                         if (!$webfinger && (strstr($uri, "@"))) {
 290                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 291                                 $webfinger = self::webfinger($path, $type);
 292                         }
 293
 294                         // Special treatment for Mastodon
 295                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 296                         // But the webfinger for this format fails.
 297                         if (!$webfinger && !empty($nick)) {
 298                                 // Mastodon uses a "@" as prefix for usernames in their url format
 299                                 $nick = ltrim($nick, '@');
 300
 301                                 $addr = $nick."@".$host;
 302
 303                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 304                                 $webfinger = self::webfinger($path, $type);
 305                         }
 306                 }
 307
 308                 if (!is_array($webfinger["links"])) {
 309                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
 310                         return false;
 311                 }
 312
 313                 $data = [];
 314
 315                 foreach ($webfinger["links"] as $link) {
 316                         $data[] = ["@attributes" => $link];
 317                 }
 318
 319                 if (is_array($webfinger["aliases"])) {
 320                         foreach ($webfinger["aliases"] as $alias) {
 321                                 $data[] = ["@attributes" =>
 322                                                         ["rel" => "alias",
 323                                                                 "href" => $alias]];
 324                         }
 325                 }
 326
 327                 return $data;
 328         }
 329
 330         /**
 331          * Fetch information (protocol endpoints and user information) about a given uri
 332          *
 333          * @param string  $uri     Address that should be probed
 334          * @param string  $network Test for this specific network
 335          * @param integer $uid     User ID for the probe (only used for mails)
 336          * @param boolean $cache   Use cached values?
 337          *
 338          * @return array uri data
 339          * @throws HTTPException\InternalServerErrorException
 340          * @throws \ImagickException
 341          */
 342         public static function uri($uri, $network = '', $uid = -1, $cache = true)
 343         {
 344                 if ($cache) {
 345                         $result = DI::cache()->get('Probe::uri:' . $network . ':' . $uri);
 346                         if (!is_null($result)) {
 347                                 return $result;
 348                         }
 349                 }
 350
 351                 if ($uid == -1) {
 352                         $uid = local_user();
 353                 }
 354
 355                 self::$istimeout = false;
 356
 357                 if ($network != Protocol::ACTIVITYPUB) {
 358                         $data = self::detect($uri, $network, $uid);
 359                 } else {
 360                         $data = null;
 361                 }
 362
 363                 // When the previous detection process had got a time out
 364                 // we could falsely detect a Friendica profile as AP profile.
 365                 if (!self::$istimeout) {
 366                         $ap_profile = ActivityPub::probeProfile($uri);
 367
 368                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 369                                 $data = $ap_profile;
 370                         } elseif (!empty($ap_profile)) {
 371                                 $ap_profile['batch'] = '';
 372                                 $data = array_merge($ap_profile, $data);
 373                         }
 374                 } else {
 375                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 376                 }
 377
 378                 if (!isset($data['url'])) {
 379                         $data['url'] = $uri;
 380                 }
 381
 382                 if (!empty($data['photo']) && !empty($data['baseurl'])) {
 383                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl']), Strings::normaliseLink($data['photo']));
 384                 } elseif (empty($data['photo'])) {
 385                         $data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
 386                 }
 387
 388                 if (empty($data['name'])) {
 389                         if (!empty($data['nick'])) {
 390                                 $data['name'] = $data['nick'];
 391                         }
 392
 393                         if (empty($data['name'])) {
 394                                 $data['name'] = $data['url'];
 395                         }
 396                 }
 397
 398                 if (empty($data['nick'])) {
 399                         $data['nick'] = strtolower($data['name']);
 400
 401                         if (strpos($data['nick'], ' ')) {
 402                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 403                         }
 404                 }
 405
 406                 if (!empty(self::$baseurl)) {
 407                         $data['baseurl'] = self::$baseurl;
 408                 }
 409
 410                 if (empty($data['network'])) {
 411                         $data['network'] = Protocol::PHANTOM;
 412                 }
 413
 414                 // Ensure that local connections always are DFRN
 415                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 416                         $data['network'] = Protocol::DFRN;
 417                 }
 418
 419                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 420                         $data['hide'] = self::getHideStatus($data['url']);
 421                 }
 422
 423                 $data = self::rearrangeData($data);
 424
 425                 // Only store into the cache if the value seems to be valid
 426                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 427                         DI::cache()->set('Probe::uri:' . $network . ':' . $uri, $data, Duration::DAY);
 428                 }
 429
 430                 return $data;
 431         }
 432
 433
 434         /**
 435          * Fetches the "hide" status from the profile
 436          *
 437          * @param string $url URL of the profile
 438          *
 439          * @return boolean "hide" status
 440          */
 441         private static function getHideStatus($url)
 442         {
 443                 $curlResult = Network::curl($url);
 444                 if (!$curlResult->isSuccess()) {
 445                         return false;
 446                 }
 447
 448                 // If the file is too large then exit
 449                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 450                         return false;
 451                 }
 452
 453                 // If it isn't a HTML file then exit
 454                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 455                         return false;
 456                 }
 457
 458                 $body = $curlResult->getBody();
 459
 460                 $doc = new DOMDocument();
 461                 @$doc->loadHTML($body);
 462
 463                 $xpath = new DOMXPath($doc);
 464
 465                 $list = $xpath->query('//meta[@name]');
 466                 foreach ($list as $node) {
 467                         $meta_tag = [];
 468                         if ($node->attributes->length) {
 469                                 foreach ($node->attributes as $attribute) {
 470                                         $meta_tag[$attribute->name] = $attribute->value;
 471                                 }
 472                         }
 473
 474                         if (empty($meta_tag['content'])) {
 475                                 continue;
 476                         }
 477
 478                         $content = strtolower(trim($meta_tag['content']));
 479
 480                         switch (strtolower(trim($meta_tag['name']))) {
 481                                 case 'dfrn-global-visibility':
 482                                         if ($content == 'false') {
 483                                                 return true;
 484                                         }
 485                                         break;
 486                                 case 'robots':
 487                                         if (strpos($content, 'noindex') !== false) {
 488                                                 return true;
 489                                         }
 490                                         break;
 491                         }
 492                 }
 493
 494                 return false;
 495         }
 496
 497         /**
 498          * Checks if a profile url should be OStatus but only provides partial information
 499          *
 500          * @param array  $webfinger Webfinger data
 501          * @param string $lrdd      Path template for webfinger request
 502          * @param string $type      type
 503          *
 504          * @return array fixed webfinger data
 505          * @throws HTTPException\InternalServerErrorException
 506          */
 507         private static function fixOStatus($webfinger, $lrdd, $type)
 508         {
 509                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 510                         return $webfinger;
 511                 }
 512
 513                 $is_ostatus = false;
 514                 $has_key = false;
 515
 516                 foreach ($webfinger['links'] as $link) {
 517                         if ($link['rel'] == ActivityNamespace::OSTATUSSUB) {
 518                                 $is_ostatus = true;
 519                         }
 520                         if ($link['rel'] == 'magic-public-key') {
 521                                 $has_key = true;
 522                         }
 523                 }
 524
 525                 if (!$is_ostatus || $has_key) {
 526                         return $webfinger;
 527                 }
 528
 529                 $url = Network::switchScheme($webfinger['subject']);
 530                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 531                 $webfinger2 = self::webfinger($path, $type);
 532
 533                 // Is the new webfinger detectable as OStatus?
 534                 if (self::ostatus($webfinger2, true)) {
 535                         $webfinger = $webfinger2;
 536                 }
 537
 538                 return $webfinger;
 539         }
 540
 541         /**
 542          * Fetch information (protocol endpoints and user information) about a given uri
 543          *
 544          * This function is only called by the "uri" function that adds caching and rearranging of data.
 545          *
 546          * @param string  $uri     Address that should be probed
 547          * @param string  $network Test for this specific network
 548          * @param integer $uid     User ID for the probe (only used for mails)
 549          *
 550          * @return array uri data
 551          * @throws HTTPException\InternalServerErrorException
 552          */
 553         private static function detect($uri, $network, $uid)
 554         {
 555                 $parts = parse_url($uri);
 556
 557                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 558                         $host = $parts["host"];
 559                         if (!empty($parts["port"])) {
 560                                 $host .= ':'.$parts["port"];
 561                         }
 562
 563                         if ($host == 'twitter.com') {
 564                                 return self::twitter($uri);
 565                         }
 566                         $lrdd = self::hostMeta($host);
 567
 568                         if (is_bool($lrdd)) {
 569                                 return [];
 570                         }
 571
 572                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
 573
 574                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 575                                 $host .= "/".array_shift($path_parts);
 576                                 $lrdd = self::hostMeta($host);
 577                         }
 578                         if (!$lrdd) {
 579                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 580                                 return self::feed($uri);
 581                         }
 582                         $nick = array_pop($path_parts);
 583
 584                         // Mastodon uses a "@" as prefix for usernames in their url format
 585                         $nick = ltrim($nick, '@');
 586
 587                         $addr = $nick."@".$host;
 588                 } elseif (strstr($uri, '@')) {
 589                         // If the URI starts with "mailto:" then jump directly to the mail detection
 590                         if (strpos($uri, 'mailto:') !== false) {
 591                                 $uri = str_replace('mailto:', '', $uri);
 592                                 return self::mail($uri, $uid);
 593                         }
 594
 595                         if ($network == Protocol::MAIL) {
 596                                 return self::mail($uri, $uid);
 597                         }
 598                         // Remove "acct:" from the URI
 599                         $uri = str_replace('acct:', '', $uri);
 600
 601                         $host = substr($uri, strpos($uri, '@') + 1);
 602                         $nick = substr($uri, 0, strpos($uri, '@'));
 603
 604                         if (strpos($uri, '@twitter.com')) {
 605                                 return self::twitter($uri);
 606                         }
 607                         $lrdd = self::hostMeta($host);
 608
 609                         if (is_bool($lrdd)) {
 610                                 return [];
 611                         }
 612
 613                         if (!$lrdd) {
 614                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 615                                 return self::mail($uri, $uid);
 616                         }
 617                         $addr = $uri;
 618                 } else {
 619                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 620                         return false;
 621                 }
 622
 623                 $webfinger = false;
 624
 625                 /// @todo Do we need the prefix "acct:" or "acct://"?
 626
 627                 foreach ($lrdd as $type => $template) {
 628                         if ($webfinger) {
 629                                 continue;
 630                         }
 631
 632                         // At first try it with the given uri
 633                         $path = str_replace('{uri}', urlencode($uri), $template);
 634                         $webfinger = self::webfinger($path, $type);
 635
 636                         // Fix possible problems with GNU Social probing to wrong scheme
 637                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 638
 639                         // We cannot be sure that the detected address was correct, so we don't use the values
 640                         if ($webfinger && ($uri != $addr)) {
 641                                 $nick = "";
 642                                 $addr = "";
 643                         }
 644
 645                         // Try webfinger with the address (user@domain.tld)
 646                         if (!$webfinger) {
 647                                 $path = str_replace('{uri}', urlencode($addr), $template);
 648                                 $webfinger = self::webfinger($path, $type);
 649                         }
 650
 651                         // Mastodon needs to have it with "acct:"
 652                         if (!$webfinger) {
 653                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 654                                 $webfinger = self::webfinger($path, $type);
 655                         }
 656                 }
 657
 658                 if (!$webfinger) {
 659                         return self::feed($uri);
 660                 }
 661
 662                 $result = false;
 663
 664                 Logger::log("Probing ".$uri, Logger::DEBUG);
 665
 666                 if (in_array($network, ["", Protocol::DFRN])) {
 667                         $result = self::dfrn($webfinger);
 668                 }
 669                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 670                         $result = self::diaspora($webfinger);
 671                 }
 672                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 673                         $result = self::ostatus($webfinger);
 674                 }
 675                 if (in_array($network, ['', Protocol::ZOT])) {
 676                         $result = self::zot($webfinger, $result);
 677                 }
 678                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 679                         $result = self::pumpio($webfinger, $addr);
 680                 }
 681                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 682                         $result = self::feed($uri);
 683                 } else {
 684                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 685                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 686                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 687                                 $result["nick"] = $nick;
 688                         }
 689
 690                         if (empty($result["addr"]) && ($addr != "")) {
 691                                 $result["addr"] = $addr;
 692                         }
 693                 }
 694
 695                 if (empty($result["network"])) {
 696                         $result["network"] = Protocol::PHANTOM;
 697                 }
 698
 699                 if (empty($result["url"])) {
 700                         $result["url"] = $uri;
 701                 }
 702
 703                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 704
 705                 if (empty($result["baseurl"]) && ($result["network"] != Protocol::PHANTOM)) {
 706                         $pos = strpos($result["url"], $host);
 707                         if ($pos) {
 708                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 709                         }
 710                 }
 711                 return $result;
 712         }
 713
 714         /**
 715          * Check for Zot contact
 716          *
 717          * @param array $webfinger Webfinger data
 718          * @param array $data      previously probed data
 719          *
 720          * @return array Zot data
 721          * @throws HTTPException\InternalServerErrorException
 722          */
 723         private static function zot($webfinger, $data)
 724         {
 725                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 726                         foreach ($webfinger["aliases"] as $alias) {
 727                                 if (substr($alias, 0, 5) == 'acct:') {
 728                                         $data["addr"] = substr($alias, 5);
 729                                 }
 730                         }
 731                 }
 732
 733                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 734                         $data["addr"] = substr($webfinger["subject"], 5);
 735                 }
 736
 737                 $zot_url = '';
 738                 foreach ($webfinger['links'] as $link) {
 739                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 740                                 $zot_url = $link['href'];
 741                         }
 742                 }
 743
 744                 if (empty($zot_url) && !empty($data['addr']) && !empty(self::$baseurl)) {
 745                         $condition = ['nurl' => Strings::normaliseLink(self::$baseurl), 'platform' => ['hubzilla']];
 746                         if (!DBA::exists('gserver', $condition)) {
 747                                 return $data;
 748                         }
 749                         $zot_url = self::$baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 750                 }
 751
 752                 if (empty($zot_url)) {
 753                         return $data;
 754                 }
 755
 756                 $data = self::pollZot($zot_url, $data);
 757
 758                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 759                         foreach ($webfinger['aliases'] as $alias) {
 760                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 761                                         $data['alias'] = $alias;
 762                                 }
 763                         }
 764                 }
 765
 766                 return $data;
 767         }
 768
 769         public static function pollZot($url, $data)
 770         {
 771                 $curlResult = Network::curl($url);
 772                 if ($curlResult->isTimeout()) {
 773                         return $data;
 774                 }
 775                 $content = $curlResult->getBody();
 776                 if (!$content) {
 777                         return $data;
 778                 }
 779
 780                 $json = json_decode($content, true);
 781                 if (!is_array($json)) {
 782                         return $data;
 783                 }
 784
 785                 if (empty($data['network'])) {
 786                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 787                                 $data['network'] = Protocol::ZOT;
 788                         } elseif (!isset($json['protocols'])) {
 789                                 $data['network'] = Protocol::ZOT;
 790                         }
 791                 }
 792
 793                 if (!empty($json['guid']) && empty($data['guid'])) {
 794                         $data['guid'] = $json['guid'];
 795                 }
 796                 if (!empty($json['key']) && empty($data['pubkey'])) {
 797                         $data['pubkey'] = $json['key'];
 798                 }
 799                 if (!empty($json['name'])) {
 800                         $data['name'] = $json['name'];
 801                 }
 802                 if (!empty($json['photo'])) {
 803                         $data['photo'] = $json['photo'];
 804                         if (!empty($json['photo_updated'])) {
 805                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 806                         }
 807                 }
 808                 if (!empty($json['address'])) {
 809                         $data['addr'] = $json['address'];
 810                 }
 811                 if (!empty($json['url'])) {
 812                         $data['url'] = $json['url'];
 813                 }
 814                 if (!empty($json['connections_url'])) {
 815                         $data['poco'] = $json['connections_url'];
 816                 }
 817                 if (isset($json['searchable'])) {
 818                         $data['hide'] = !$json['searchable'];
 819                 }
 820                 if (!empty($json['public_forum'])) {
 821                         $data['community'] = $json['public_forum'];
 822                         $data['account-type'] = Contact::PAGE_COMMUNITY;
 823                 }
 824
 825                 if (!empty($json['profile'])) {
 826                         $profile = $json['profile'];
 827                         if (!empty($profile['description'])) {
 828                                 $data['about'] = $profile['description'];
 829                         }
 830                         if (!empty($profile['gender'])) {
 831                                 $data['gender'] = $profile['gender'];
 832                         }
 833                         if (!empty($profile['keywords'])) {
 834                                 $keywords = implode(', ', $profile['keywords']);
 835                                 if (!empty($keywords)) {
 836                                         $data['keywords'] = $keywords;
 837                                 }
 838                         }
 839
 840                         $loc = [];
 841                         if (!empty($profile['region'])) {
 842                                 $loc['region'] = $profile['region'];
 843                         }
 844                         if (!empty($profile['country'])) {
 845                                 $loc['country-name'] = $profile['country'];
 846                         }
 847                         $location = Profile::formatLocation($loc);
 848                         if (!empty($location)) {
 849                                 $data['location'] = $location;
 850                         }
 851                 }
 852
 853                 return $data;
 854         }
 855
 856         /**
 857          * Perform a webfinger request.
 858          *
 859          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 860          *
 861          * @param string $url  Address that should be probed
 862          * @param string $type type
 863          *
 864          * @return array webfinger data
 865          * @throws HTTPException\InternalServerErrorException
 866          */
 867         private static function webfinger($url, $type)
 868         {
 869                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 870
 871                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 872                 if ($curlResult->isTimeout()) {
 873                         self::$istimeout = true;
 874                         return false;
 875                 }
 876                 $data = $curlResult->getBody();
 877
 878                 $webfinger = json_decode($data, true);
 879                 if (is_array($webfinger)) {
 880                         if (!isset($webfinger["links"])) {
 881                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 882                                 return false;
 883                         }
 884                         return $webfinger;
 885                 }
 886
 887                 // If it is not JSON, maybe it is XML
 888                 $xrd = XML::parseString($data, false);
 889                 if (!is_object($xrd)) {
 890                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 891                         return false;
 892                 }
 893
 894                 $xrd_arr = XML::elementToArray($xrd);
 895                 if (!isset($xrd_arr["xrd"]["link"])) {
 896                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 897                         return false;
 898                 }
 899
 900                 $webfinger = [];
 901
 902                 if (!empty($xrd_arr["xrd"]["subject"])) {
 903                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 904                 }
 905
 906                 if (!empty($xrd_arr["xrd"]["alias"])) {
 907                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 908                 }
 909
 910                 $webfinger["links"] = [];
 911
 912                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 913                         if (!empty($data["@attributes"])) {
 914                                 $attributes = $data["@attributes"];
 915                         } elseif ($value == "@attributes") {
 916                                 $attributes = $data;
 917                         } else {
 918                                 continue;
 919                         }
 920
 921                         $webfinger["links"][] = $attributes;
 922                 }
 923                 return $webfinger;
 924         }
 925
 926         /**
 927          * Poll the Friendica specific noscrape page.
 928          *
 929          * "noscrape" is a faster alternative to fetch the data from the hcard.
 930          * This functionality was originally created for the directory.
 931          *
 932          * @param string $noscrape_url Link to the noscrape page
 933          * @param array  $data         The already fetched data
 934          *
 935          * @return array noscrape data
 936          * @throws HTTPException\InternalServerErrorException
 937          */
 938         private static function pollNoscrape($noscrape_url, $data)
 939         {
 940                 $curlResult = Network::curl($noscrape_url);
 941                 if ($curlResult->isTimeout()) {
 942                         self::$istimeout = true;
 943                         return false;
 944                 }
 945                 $content = $curlResult->getBody();
 946                 if (!$content) {
 947                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 948                         return false;
 949                 }
 950
 951                 $json = json_decode($content, true);
 952                 if (!is_array($json)) {
 953                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 954                         return false;
 955                 }
 956
 957                 if (!empty($json["fn"])) {
 958                         $data["name"] = $json["fn"];
 959                 }
 960
 961                 if (!empty($json["addr"])) {
 962                         $data["addr"] = $json["addr"];
 963                 }
 964
 965                 if (!empty($json["nick"])) {
 966                         $data["nick"] = $json["nick"];
 967                 }
 968
 969                 if (!empty($json["guid"])) {
 970                         $data["guid"] = $json["guid"];
 971                 }
 972
 973                 if (!empty($json["comm"])) {
 974                         $data["community"] = $json["comm"];
 975                 }
 976
 977                 if (!empty($json["tags"])) {
 978                         $keywords = implode(", ", $json["tags"]);
 979                         if ($keywords != "") {
 980                                 $data["keywords"] = $keywords;
 981                         }
 982                 }
 983
 984                 $location = Profile::formatLocation($json);
 985                 if ($location) {
 986                         $data["location"] = $location;
 987                 }
 988
 989                 if (!empty($json["about"])) {
 990                         $data["about"] = $json["about"];
 991                 }
 992
 993                 if (!empty($json["gender"])) {
 994                         $data["gender"] = $json["gender"];
 995                 }
 996
 997                 if (!empty($json["key"])) {
 998                         $data["pubkey"] = $json["key"];
 999                 }
1000
1001                 if (!empty($json["photo"])) {
1002                         $data["photo"] = $json["photo"];
1003                 }
1004
1005                 if (!empty($json["dfrn-request"])) {
1006                         $data["request"] = $json["dfrn-request"];
1007                 }
1008
1009                 if (!empty($json["dfrn-confirm"])) {
1010                         $data["confirm"] = $json["dfrn-confirm"];
1011                 }
1012
1013                 if (!empty($json["dfrn-notify"])) {
1014                         $data["notify"] = $json["dfrn-notify"];
1015                 }
1016
1017                 if (!empty($json["dfrn-poll"])) {
1018                         $data["poll"] = $json["dfrn-poll"];
1019                 }
1020
1021                 if (isset($json["hide"])) {
1022                         $data["hide"] = (bool)$json["hide"];
1023                 } else {
1024                         $data["hide"] = false;
1025                 }
1026
1027                 return $data;
1028         }
1029
1030         /**
1031          * Check for valid DFRN data
1032          *
1033          * @param array $data DFRN data
1034          *
1035          * @return int Number of errors
1036          */
1037         public static function validDfrn($data)
1038         {
1039                 $errors = 0;
1040                 if (!isset($data['key'])) {
1041                         $errors ++;
1042                 }
1043                 if (!isset($data['dfrn-request'])) {
1044                         $errors ++;
1045                 }
1046                 if (!isset($data['dfrn-confirm'])) {
1047                         $errors ++;
1048                 }
1049                 if (!isset($data['dfrn-notify'])) {
1050                         $errors ++;
1051                 }
1052                 if (!isset($data['dfrn-poll'])) {
1053                         $errors ++;
1054                 }
1055                 return $errors;
1056         }
1057
1058         /**
1059          * Fetch data from a DFRN profile page and via "noscrape"
1060          *
1061          * @param string $profile_link Link to the profile page
1062          *
1063          * @return array profile data
1064          * @throws HTTPException\InternalServerErrorException
1065          * @throws \ImagickException
1066          */
1067         public static function profile($profile_link)
1068         {
1069                 $data = [];
1070
1071                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
1072
1073                 // Fetch data via noscrape - this is faster
1074                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1075                 $data = self::pollNoscrape($noscrape_url, $data);
1076
1077                 if (!isset($data["notify"])
1078                         || !isset($data["confirm"])
1079                         || !isset($data["request"])
1080                         || !isset($data["poll"])
1081                         || !isset($data["name"])
1082                         || !isset($data["photo"])
1083                 ) {
1084                         $data = self::pollHcard($profile_link, $data, true);
1085                 }
1086
1087                 $prof_data = [];
1088
1089                 if (empty($data["addr"]) || empty($data["nick"])) {
1090                         $probe_data = self::uri($profile_link);
1091                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1092                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1093                 }
1094
1095                 $prof_data["addr"]         = $data["addr"];
1096                 $prof_data["nick"]         = $data["nick"];
1097                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1098                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1099                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1100                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1101                 $prof_data["photo"]        = $data['photo']   ?? null;
1102                 $prof_data["fn"]           = $data['name']    ?? null;
1103                 $prof_data["key"]          = $data['pubkey']  ?? null;
1104
1105                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
1106
1107                 return $prof_data;
1108         }
1109
1110         /**
1111          * Check for DFRN contact
1112          *
1113          * @param array $webfinger Webfinger data
1114          *
1115          * @return array DFRN data
1116          * @throws HTTPException\InternalServerErrorException
1117          */
1118         private static function dfrn($webfinger)
1119         {
1120                 $hcard_url = "";
1121                 $data = [];
1122                 // The array is reversed to take into account the order of preference for same-rel links
1123                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1124                 foreach (array_reverse($webfinger["links"]) as $link) {
1125                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1126                                 $data["network"] = Protocol::DFRN;
1127                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1128                                 $data["poll"] = $link["href"];
1129                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1130                                 $data["url"] = $link["href"];
1131                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1132                                 $hcard_url = $link["href"];
1133                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1134                                 $data["poco"] = $link["href"];
1135                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1136                                 $data["photo"] = $link["href"];
1137                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1138                                 $data["baseurl"] = trim($link["href"], '/');
1139                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1140                                 $data["guid"] = $link["href"];
1141                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1142                                 $data["pubkey"] = base64_decode($link["href"]);
1143
1144                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1145                                 if (strstr($data["pubkey"], 'RSA ')) {
1146                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1147                                 }
1148                         }
1149                 }
1150
1151                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1152                         foreach ($webfinger["aliases"] as $alias) {
1153                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1154                                         $data["url"] = $alias;
1155                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1156                                         $data["alias"] = $alias;
1157                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1158                                         $data["addr"] = substr($alias, 5);
1159                                 }
1160                         }
1161                 }
1162
1163                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1164                         $data["addr"] = substr($webfinger["subject"], 5);
1165                 }
1166
1167                 if (!isset($data["network"]) || ($hcard_url == "")) {
1168                         return false;
1169                 }
1170
1171                 // Fetch data via noscrape - this is faster
1172                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1173                 $data = self::pollNoscrape($noscrape_url, $data);
1174
1175                 if (isset($data["notify"])
1176                         && isset($data["confirm"])
1177                         && isset($data["request"])
1178                         && isset($data["poll"])
1179                         && isset($data["name"])
1180                         && isset($data["photo"])
1181                 ) {
1182                         return $data;
1183                 }
1184
1185                 $data = self::pollHcard($hcard_url, $data, true);
1186
1187                 return $data;
1188         }
1189
1190         /**
1191          * Poll the hcard page (Diaspora and Friendica specific)
1192          *
1193          * @param string  $hcard_url Link to the hcard page
1194          * @param array   $data      The already fetched data
1195          * @param boolean $dfrn      Poll DFRN specific data
1196          *
1197          * @return array hcard data
1198          * @throws HTTPException\InternalServerErrorException
1199          */
1200         private static function pollHcard($hcard_url, $data, $dfrn = false)
1201         {
1202                 $curlResult = Network::curl($hcard_url);
1203                 if ($curlResult->isTimeout()) {
1204                         self::$istimeout = true;
1205                         return false;
1206                 }
1207                 $content = $curlResult->getBody();
1208                 if (!$content) {
1209                         return false;
1210                 }
1211
1212                 $doc = new DOMDocument();
1213                 if (!@$doc->loadHTML($content)) {
1214                         return false;
1215                 }
1216
1217                 $xpath = new DomXPath($doc);
1218
1219                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1220                 if (!is_object($vcards)) {
1221                         return false;
1222                 }
1223
1224                 if (!isset($data["baseurl"])) {
1225                         $data["baseurl"] = "";
1226                 }
1227
1228                 if ($vcards->length > 0) {
1229                         $vcard = $vcards->item(0);
1230
1231                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1232                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1233                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1234                         if (($search->length > 0) && empty($data["guid"])) {
1235                                 $data["guid"] = $search->item(0)->nodeValue;
1236                         }
1237
1238                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1239                         if ($search->length > 0) {
1240                                 $data["nick"] = $search->item(0)->nodeValue;
1241                         }
1242
1243                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1244                         if ($search->length > 0) {
1245                                 $data["name"] = $search->item(0)->nodeValue;
1246                         }
1247
1248                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1249                         if ($search->length > 0) {
1250                                 $data["searchable"] = $search->item(0)->nodeValue;
1251                         }
1252
1253                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1254                         if ($search->length > 0) {
1255                                 $data["pubkey"] = $search->item(0)->nodeValue;
1256                                 if (strstr($data["pubkey"], 'RSA ')) {
1257                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1258                                 }
1259                         }
1260
1261                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1262                         if ($search->length > 0) {
1263                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1264                         }
1265                 }
1266
1267                 $avatar = [];
1268                 if (!empty($vcard)) {
1269                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1270                         foreach ($photos as $photo) {
1271                                 $attr = [];
1272                                 foreach ($photo->attributes as $attribute) {
1273                                         $attr[$attribute->name] = trim($attribute->value);
1274                                 }
1275
1276                                 if (isset($attr["src"]) && isset($attr["width"])) {
1277                                         $avatar[$attr["width"]] = $attr["src"];
1278                                 }
1279
1280                                 // We don't have a width. So we just take everything that we got.
1281                                 // This is a Hubzilla workaround which doesn't send a width.
1282                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1283                                         $avatar[] = $attr["src"];
1284                                 }
1285                         }
1286                 }
1287
1288                 if (sizeof($avatar)) {
1289                         ksort($avatar);
1290                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1291                 }
1292
1293                 if ($dfrn) {
1294                         // Poll DFRN specific data
1295                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1296                         if ($search->length > 0) {
1297                                 foreach ($search as $link) {
1298                                         //$data["request"] = $search->item(0)->nodeValue;
1299                                         $attr = [];
1300                                         foreach ($link->attributes as $attribute) {
1301                                                 $attr[$attribute->name] = trim($attribute->value);
1302                                         }
1303
1304                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1305                                 }
1306                         }
1307
1308                         // Older Friendica versions had used the "uid" field differently than newer versions
1309                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1310                                 unset($data["guid"]);
1311                         }
1312                 }
1313
1314
1315                 return $data;
1316         }
1317
1318         /**
1319          * Check for Diaspora contact
1320          *
1321          * @param array $webfinger Webfinger data
1322          *
1323          * @return array Diaspora data
1324          * @throws HTTPException\InternalServerErrorException
1325          */
1326         private static function diaspora($webfinger)
1327         {
1328                 $hcard_url = "";
1329                 $data = [];
1330
1331                 // The array is reversed to take into account the order of preference for same-rel links
1332                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1333                 foreach (array_reverse($webfinger["links"]) as $link) {
1334                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1335                                 $hcard_url = $link["href"];
1336                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1337                                 $data["baseurl"] = trim($link["href"], '/');
1338                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1339                                 $data["guid"] = $link["href"];
1340                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1341                                 $data["url"] = $link["href"];
1342                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1343                                 $data["poll"] = $link["href"];
1344                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1345                                 $data["poco"] = $link["href"];
1346                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1347                                 $data["notify"] = $link["href"];
1348                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1349                                 $data["pubkey"] = base64_decode($link["href"]);
1350
1351                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1352                                 if (strstr($data["pubkey"], 'RSA ')) {
1353                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1354                                 }
1355                         }
1356                 }
1357
1358                 if (empty($data["url"]) || empty($hcard_url)) {
1359                         return false;
1360                 }
1361
1362                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1363                         foreach ($webfinger["aliases"] as $alias) {
1364                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1365                                         $data["alias"] = $alias;
1366                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1367                                         $data["addr"] = substr($alias, 5);
1368                                 }
1369                         }
1370                 }
1371
1372                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1373                         $data["addr"] = substr($webfinger["subject"], 5);
1374                 }
1375
1376                 // Fetch further information from the hcard
1377                 $data = self::pollHcard($hcard_url, $data);
1378
1379                 if (!$data) {
1380                         return false;
1381                 }
1382
1383                 if (!empty($data["url"])
1384                         && !empty($data["guid"])
1385                         && !empty($data["baseurl"])
1386                         && !empty($data["pubkey"])
1387                         && !empty($hcard_url)
1388                 ) {
1389                         $data["network"] = Protocol::DIASPORA;
1390
1391                         // The Diaspora handle must always be lowercase
1392                         if (!empty($data["addr"])) {
1393                                 $data["addr"] = strtolower($data["addr"]);
1394                         }
1395
1396                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1397                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1398                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1399                 } else {
1400                         return false;
1401                 }
1402
1403                 return $data;
1404         }
1405
1406         /**
1407          * Check for OStatus contact
1408          *
1409          * @param array $webfinger Webfinger data
1410          * @param bool  $short     Short detection mode
1411          *
1412          * @return array|bool OStatus data or "false" on error or "true" on short mode
1413          * @throws HTTPException\InternalServerErrorException
1414          */
1415         private static function ostatus($webfinger, $short = false)
1416         {
1417                 $data = [];
1418
1419                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1420                         foreach ($webfinger["aliases"] as $alias) {
1421                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1422                                         $data["addr"] = str_replace('acct:', '', $alias);
1423                                 }
1424                         }
1425                 }
1426
1427                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1428                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1429                 ) {
1430                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1431                 }
1432
1433                 if (is_array($webfinger["links"])) {
1434                         // The array is reversed to take into account the order of preference for same-rel links
1435                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1436                         foreach (array_reverse($webfinger["links"]) as $link) {
1437                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1438                                         && (($link["type"] ?? "") == "text/html")
1439                                         && ($link["href"] != "")
1440                                 ) {
1441                                         $data["url"] = $link["href"];
1442                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1443                                         $data["notify"] = $link["href"];
1444                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1445                                         $data["poll"] = $link["href"];
1446                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1447                                         $pubkey = $link["href"];
1448
1449                                         if (substr($pubkey, 0, 5) === 'data:') {
1450                                                 if (strstr($pubkey, ',')) {
1451                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1452                                                 } else {
1453                                                         $pubkey = substr($pubkey, 5);
1454                                                 }
1455                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1456                                                 $curlResult = Network::curl($pubkey);
1457                                                 if ($curlResult->isTimeout()) {
1458                                                         self::$istimeout = true;
1459                                                         return false;
1460                                                 }
1461                                                 $pubkey = $curlResult->getBody();
1462                                         }
1463
1464                                         $key = explode(".", $pubkey);
1465
1466                                         if (sizeof($key) >= 3) {
1467                                                 $m = Strings::base64UrlDecode($key[1]);
1468                                                 $e = Strings::base64UrlDecode($key[2]);
1469                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1470                                         }
1471                                 }
1472                         }
1473                 }
1474
1475                 if (isset($data["notify"]) && isset($data["pubkey"])
1476                         && isset($data["poll"])
1477                         && isset($data["url"])
1478                 ) {
1479                         $data["network"] = Protocol::OSTATUS;
1480                 } else {
1481                         return false;
1482                 }
1483
1484                 if ($short) {
1485                         return true;
1486                 }
1487
1488                 // Fetch all additional data from the feed
1489                 $curlResult = Network::curl($data["poll"]);
1490                 if ($curlResult->isTimeout()) {
1491                         self::$istimeout = true;
1492                         return false;
1493                 }
1494                 $feed = $curlResult->getBody();
1495                 $feed_data = Feed::import($feed);
1496                 if (!$feed_data) {
1497                         return false;
1498                 }
1499
1500                 if (!empty($feed_data["header"]["author-name"])) {
1501                         $data["name"] = $feed_data["header"]["author-name"];
1502                 }
1503                 if (!empty($feed_data["header"]["author-nick"])) {
1504                         $data["nick"] = $feed_data["header"]["author-nick"];
1505                 }
1506                 if (!empty($feed_data["header"]["author-avatar"])) {
1507                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1508                 }
1509                 if (!empty($feed_data["header"]["author-id"])) {
1510                         $data["alias"] = $feed_data["header"]["author-id"];
1511                 }
1512                 if (!empty($feed_data["header"]["author-location"])) {
1513                         $data["location"] = $feed_data["header"]["author-location"];
1514                 }
1515                 if (!empty($feed_data["header"]["author-about"])) {
1516                         $data["about"] = $feed_data["header"]["author-about"];
1517                 }
1518                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1519                 // So we take the value that we just fetched, although the other one worked as well
1520                 if (!empty($feed_data["header"]["author-link"])) {
1521                         $data["url"] = $feed_data["header"]["author-link"];
1522                 }
1523
1524                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1525                         $data['url'] = $data["alias"];
1526                         $data["alias"] = '';
1527                 }
1528
1529                 /// @todo Fetch location and "about" from the feed as well
1530                 return $data;
1531         }
1532
1533         /**
1534          * Fetch data from a pump.io profile page
1535          *
1536          * @param string $profile_link Link to the profile page
1537          *
1538          * @return array profile data
1539          */
1540         private static function pumpioProfileData($profile_link)
1541         {
1542                 $curlResult = Network::curl($profile_link);
1543                 if (!$curlResult->isSuccess()) {
1544                         return false;
1545                 }
1546
1547                 $doc = new DOMDocument();
1548                 if (!@$doc->loadHTML($curlResult->getBody())) {
1549                         return false;
1550                 }
1551
1552                 $xpath = new DomXPath($doc);
1553
1554                 $data = [];
1555
1556                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1557
1558                 if ($data["name"] == '') {
1559                         // This is ugly - but pump.io doesn't seem to know a better way for it
1560                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1561                         $pos = strpos($data["name"], chr(10));
1562                         if ($pos) {
1563                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1564                         }
1565                 }
1566
1567                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1568
1569                 if ($data["location"] == '') {
1570                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1571                 }
1572
1573                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1574
1575                 if ($data["about"] == '') {
1576                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1577                 }
1578
1579                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1580                 if (!$avatar) {
1581                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1582                 }
1583                 if ($avatar) {
1584                         foreach ($avatar->attributes as $attribute) {
1585                                 if ($attribute->name == "src") {
1586                                         $data["photo"] = trim($attribute->value);
1587                                 }
1588                         }
1589                 }
1590
1591                 return $data;
1592         }
1593
1594         /**
1595          * Check for pump.io contact
1596          *
1597          * @param array  $webfinger Webfinger data
1598          * @param string $addr
1599          * @return array pump.io data
1600          */
1601         private static function pumpio($webfinger, $addr)
1602         {
1603                 $data = [];
1604                 // The array is reversed to take into account the order of preference for same-rel links
1605                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1606                 foreach (array_reverse($webfinger["links"]) as $link) {
1607                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1608                                 && (($link["type"] ?? "") == "text/html")
1609                                 && ($link["href"] != "")
1610                         ) {
1611                                 $data["url"] = $link["href"];
1612                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1613                                 $data["notify"] = $link["href"];
1614                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1615                                 $data["poll"] = $link["href"];
1616                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1617                                 $data["dialback"] = $link["href"];
1618                         }
1619                 }
1620                 if (isset($data["poll"]) && isset($data["notify"])
1621                         && isset($data["dialback"])
1622                         && isset($data["url"])
1623                 ) {
1624                         // by now we use these fields only for the network type detection
1625                         // So we unset all data that isn't used at the moment
1626                         unset($data["dialback"]);
1627
1628                         $data["network"] = Protocol::PUMPIO;
1629                 } else {
1630                         return false;
1631                 }
1632
1633                 $profile_data = self::pumpioProfileData($data["url"]);
1634
1635                 if (!$profile_data) {
1636                         return false;
1637                 }
1638
1639                 $data = array_merge($data, $profile_data);
1640
1641                 if (($addr != '') && ($data['name'] != '')) {
1642                         $name = trim(str_replace($addr, '', $data['name']));
1643                         if ($name != '') {
1644                                 $data['name'] = $name;
1645                         }
1646                 }
1647
1648                 return $data;
1649         }
1650
1651         /**
1652          * Check for twitter contact
1653          *
1654          * @param string $uri
1655          *
1656          * @return array twitter data
1657          */
1658         private static function twitter($uri)
1659         {
1660                 if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) {
1661                         $nick = $matches[1];
1662                 } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) {
1663                         $nick = $matches[1];
1664                 } else {
1665                         return [];
1666                 }
1667
1668                 $data = [];
1669                 $data['url'] = 'https://twitter.com/' . $nick;
1670                 $data['addr'] = $nick . '@twitter.com';
1671                 $data['nick'] = $data['name'] = $nick;
1672                 $data['network'] = Protocol::TWITTER;
1673                 $data['baseurl'] = 'https://twitter.com';
1674
1675                 $curlResult = Network::curl($data['url'], false);
1676                 if (!$curlResult->isSuccess()) {
1677                         return [];
1678                 }
1679
1680                 $body = $curlResult->getBody();
1681                 $doc = new DOMDocument();
1682                 @$doc->loadHTML($body);
1683                 $xpath = new DOMXPath($doc);
1684
1685                 $list = $xpath->query('//img[@class]');
1686                 foreach ($list as $node) {
1687                         $img_attr = [];
1688                         if ($node->attributes->length) {
1689                                 foreach ($node->attributes as $attribute) {
1690                                         $img_attr[$attribute->name] = $attribute->value;
1691                                 }
1692                         }
1693
1694                         if (empty($img_attr['class'])) {
1695                                 continue;
1696                         }
1697
1698                         if (strpos($img_attr['class'], 'ProfileAvatar-image') !== false) {
1699                                 if (!empty($img_attr['src'])) {
1700                                         $data['photo'] = $img_attr['src'];
1701                                 }
1702                                 if (!empty($img_attr['alt'])) {
1703                                         $data['name'] = $img_attr['alt'];
1704                                 }
1705                         }
1706                 }
1707
1708                 return $data;
1709         }
1710
1711         /**
1712          * Check page for feed link
1713          *
1714          * @param string $url Page link
1715          *
1716          * @return string feed link
1717          */
1718         private static function getFeedLink($url)
1719         {
1720                 $curlResult = Network::curl($url);
1721                 if (!$curlResult->isSuccess()) {
1722                         return false;
1723                 }
1724
1725                 $doc = new DOMDocument();
1726                 if (!@$doc->loadHTML($curlResult->getBody())) {
1727                         return false;
1728                 }
1729
1730                 $xpath = new DomXPath($doc);
1731
1732                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1733                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1734                 if (!is_object($feeds)) {
1735                         return false;
1736                 }
1737
1738                 if ($feeds->length == 0) {
1739                         return false;
1740                 }
1741
1742                 $feed_url = "";
1743
1744                 foreach ($feeds as $feed) {
1745                         $attr = [];
1746                         foreach ($feed->attributes as $attribute) {
1747                                 $attr[$attribute->name] = trim($attribute->value);
1748                         }
1749
1750                         if (empty($feed_url) && !empty($attr['href'])) {
1751                                 $feed_url = $attr["href"];
1752                         }
1753                 }
1754
1755                 return $feed_url;
1756         }
1757
1758         /**
1759          * Check for feed contact
1760          *
1761          * @param string  $url   Profile link
1762          * @param boolean $probe Do a probe if the page contains a feed link
1763          *
1764          * @return array feed data
1765          * @throws HTTPException\InternalServerErrorException
1766          */
1767         private static function feed($url, $probe = true)
1768         {
1769                 $curlResult = Network::curl($url);
1770                 if ($curlResult->isTimeout()) {
1771                         self::$istimeout = true;
1772                         return false;
1773                 }
1774                 $feed = $curlResult->getBody();
1775                 $feed_data = Feed::import($feed);
1776
1777                 if (!$feed_data) {
1778                         if (!$probe) {
1779                                 return false;
1780                         }
1781
1782                         $feed_url = self::getFeedLink($url);
1783
1784                         if (!$feed_url) {
1785                                 return false;
1786                         }
1787
1788                         return self::feed($feed_url, false);
1789                 }
1790
1791                 if (!empty($feed_data["header"]["author-name"])) {
1792                         $data["name"] = $feed_data["header"]["author-name"];
1793                 }
1794
1795                 if (!empty($feed_data["header"]["author-nick"])) {
1796                         $data["nick"] = $feed_data["header"]["author-nick"];
1797                 }
1798
1799                 if (!empty($feed_data["header"]["author-avatar"])) {
1800                         $data["photo"] = $feed_data["header"]["author-avatar"];
1801                 }
1802
1803                 if (!empty($feed_data["header"]["author-id"])) {
1804                         $data["alias"] = $feed_data["header"]["author-id"];
1805                 }
1806
1807                 $data["url"] = $url;
1808                 $data["poll"] = $url;
1809
1810                 if (!empty($feed_data["header"]["author-link"])) {
1811                         $data["baseurl"] = $feed_data["header"]["author-link"];
1812                 } else {
1813                         $data["baseurl"] = $data["url"];
1814                 }
1815
1816                 $data["network"] = Protocol::FEED;
1817
1818                 return $data;
1819         }
1820
1821         /**
1822          * Check for mail contact
1823          *
1824          * @param string  $uri Profile link
1825          * @param integer $uid User ID
1826          *
1827          * @return array mail data
1828          * @throws \Exception
1829          */
1830         private static function mail($uri, $uid)
1831         {
1832                 if (!Network::isEmailDomainValid($uri)) {
1833                         return false;
1834                 }
1835
1836                 if ($uid == 0) {
1837                         return false;
1838                 }
1839
1840                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1841
1842                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1843                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1844                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1845
1846                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1847                         return false;
1848                 }
1849
1850                 $mailbox = Email::constructMailboxName($mailacct);
1851                 $password = '';
1852                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1853                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1854                 if (!$mbox) {
1855                         return false;
1856                 }
1857
1858                 $msgs = Email::poll($mbox, $uri);
1859                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1860
1861                 if (!count($msgs)) {
1862                         return false;
1863                 }
1864
1865                 $phost = substr($uri, strpos($uri, '@') + 1);
1866
1867                 $data = [];
1868                 $data["addr"]    = $uri;
1869                 $data["network"] = Protocol::MAIL;
1870                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1871                 $data["nick"]    = $data["name"];
1872                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1873                 $data["url"]     = 'mailto:'.$uri;
1874                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1875                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1876
1877                 $x = Email::messageMeta($mbox, $msgs[0]);
1878                 if (stristr($x[0]->from, $uri)) {
1879                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1880                 } elseif (stristr($x[0]->to, $uri)) {
1881                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1882                 }
1883                 if (isset($adr)) {
1884                         foreach ($adr as $feadr) {
1885                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1886                                         &&(strcasecmp($feadr->host, $phost) == 0)
1887                                         && (strlen($feadr->personal))
1888                                 ) {
1889                                         $personal = imap_mime_header_decode($feadr->personal);
1890                                         $data["name"] = "";
1891                                         foreach ($personal as $perspart) {
1892                                                 if ($perspart->charset != "default") {
1893                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1894                                                 } else {
1895                                                         $data["name"] .= $perspart->text;
1896                                                 }
1897                                         }
1898
1899                                         $data["name"] = Strings::escapeTags($data["name"]);
1900                                 }
1901                         }
1902                 }
1903                 if (!empty($mbox)) {
1904                         imap_close($mbox);
1905                 }
1906                 return $data;
1907         }
1908
1909         /**
1910          * Mix two paths together to possibly fix missing parts
1911          *
1912          * @param string $avatar Path to the avatar
1913          * @param string $base   Another path that is hopefully complete
1914          *
1915          * @return string fixed avatar path
1916          * @throws \Exception
1917          */
1918         public static function fixAvatar($avatar, $base)
1919         {
1920                 $base_parts = parse_url($base);
1921
1922                 // Remove all parts that could create a problem
1923                 unset($base_parts['path']);
1924                 unset($base_parts['query']);
1925                 unset($base_parts['fragment']);
1926
1927                 $avatar_parts = parse_url($avatar);
1928
1929                 // Now we mix them
1930                 $parts = array_merge($base_parts, $avatar_parts);
1931
1932                 // And put them together again
1933                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1934                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1935                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1936                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1937                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1938                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1939
1940                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1941
1942                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1943
1944                 return $fixed;
1945         }
1946 }