src/Network/Probe.php

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