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\Core\System;
  19 use Friendica\Database\DBA;
  20 use Friendica\Model\Profile;
  21 use Friendica\Protocol\ActivityPub;
  22 use Friendica\Protocol\Email;
  23 use Friendica\Protocol\Feed;
  24 use Friendica\Util\Crypto;
  25 use Friendica\Util\DateTimeFormat;
  26 use Friendica\Util\Network;
  27 use Friendica\Util\Strings;
  28 use Friendica\Util\XML;
  29
  30 /**
  31  * @brief This class contain functions for probing URL
  32  *
  33  */
  34 class Probe
  35 {
  36         private static $baseurl;
  37         private static $istimeout;
  38
  39         /**
  40          * @brief Rearrange the array so that it always has the same order
  41          *
  42          * @param array $data Unordered data
  43          *
  44          * @return array Ordered data
  45          */
  46         private static function rearrangeData($data)
  47         {
  48                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo",
  49                                 "account-type", "community", "keywords", "location", "about",
  50                                 "batch", "notify", "poll", "request", "confirm", "poco",
  51                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  52                                 "priority", "network", "pubkey", "baseurl"];
  53
  54                 $newdata = [];
  55                 foreach ($fields as $field) {
  56                         if (isset($data[$field])) {
  57                                 $newdata[$field] = $data[$field];
  58                         } else {
  59                                 $newdata[$field] = "";
  60                         }
  61                 }
  62
  63                 // We don't use the "priority" field anymore and replace it with a dummy.
  64                 $newdata["priority"] = 0;
  65
  66                 return $newdata;
  67         }
  68
  69         /**
  70          * @brief Check if the hostname belongs to the own server
  71          *
  72          * @param string $host The hostname that is to be checked
  73          *
  74          * @return bool Does the testes hostname belongs to the own server?
  75          */
  76         private static function ownHost($host)
  77         {
  78                 $own_host = \get_app()->getHostName();
  79
  80                 $parts = parse_url($host);
  81
  82                 if (!isset($parts['scheme'])) {
  83                         $parts = parse_url('http://'.$host);
  84                 }
  85
  86                 if (!isset($parts['host'])) {
  87                         return false;
  88                 }
  89                 return $parts['host'] == $own_host;
  90         }
  91
  92         /**
  93          * @brief Probes for webfinger path via "host-meta"
  94          *
  95          * We have to check if the servers in the future still will offer this.
  96          * It seems as if it was dropped from the standard.
  97          *
  98          * @param string $host The host part of an url
  99          *
 100          * @return array with template and type of the webfinger template for JSON or XML
 101          * @throws HTTPException\InternalServerErrorException
 102          */
 103         private static function hostMeta($host)
 104         {
 105                 // Reset the static variable
 106                 self::$baseurl = '';
 107
 108                 $ssl_url = "https://".$host."/.well-known/host-meta";
 109                 $url = "http://".$host."/.well-known/host-meta";
 110
 111                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 112
 113                 Logger::log("Probing for ".$host, Logger::DEBUG);
 114                 $xrd = null;
 115
 116                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 117                 if ($curlResult->isSuccess()) {
 118                         $xml = $curlResult->getBody();
 119                         $xrd = XML::parseString($xml, false);
 120                         $host_url = 'https://'.$host;
 121                 }
 122
 123                 if (!is_object($xrd)) {
 124                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 125                         if ($curlResult->isTimeout()) {
 126                                 Logger::log("Probing timeout for " . $url, Logger::DEBUG);
 127                                 self::$istimeout = true;
 128                                 return false;
 129                         }
 130                         $xml = $curlResult->getBody();
 131                         $xrd = XML::parseString($xml, false);
 132                         $host_url = 'http://'.$host;
 133                 }
 134                 if (!is_object($xrd)) {
 135                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
 136                         return [];
 137                 }
 138
 139                 $links = XML::elementToArray($xrd);
 140                 if (!isset($links["xrd"]["link"])) {
 141                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
 142                         return [];
 143                 }
 144
 145                 $lrdd = [];
 146                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 147                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 148                 // compared to the XML webfinger. So this is commented out by now.
 149                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 150
 151                 foreach ($links["xrd"]["link"] as $value => $link) {
 152                         if (!empty($link["@attributes"])) {
 153                                 $attributes = $link["@attributes"];
 154                         } elseif ($value == "@attributes") {
 155                                 $attributes = $link;
 156                         } else {
 157                                 continue;
 158                         }
 159
 160                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 161                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 162
 163                                 $lrdd[$type] = $attributes["template"];
 164                         }
 165                 }
 166
 167                 self::$baseurl = $host_url;
 168
 169                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
 170
 171                 return $lrdd;
 172         }
 173
 174         /**
 175          * @brief Perform Webfinger lookup and return DFRN data
 176          *
 177          * Given an email style address, perform webfinger lookup and
 178          * return the resulting DFRN profile URL, or if no DFRN profile URL
 179          * is located, returns an OStatus subscription template (prefixed
 180          * with the string 'stat:' to identify it as on OStatus template).
 181          * If this isn't an email style address just return $webbie.
 182          * Return an empty string if email-style addresses but webfinger fails,
 183          * or if the resultant personal XRD doesn't contain a supported
 184          * subscription/friend-request attribute.
 185          *
 186          * amended 7/9/2011 to return an hcard which could save potentially loading
 187          * a lengthy content page to scrape dfrn attributes
 188          *
 189          * @param string $webbie    Address that should be probed
 190          * @param string $hcard_url Link to the hcard - is returned by reference
 191          *
 192          * @return string profile link
 193          * @throws HTTPException\InternalServerErrorException
 194          */
 195         public static function webfingerDfrn($webbie, &$hcard_url)
 196         {
 197                 $profile_link = '';
 198
 199                 $links = self::lrdd($webbie);
 200                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
 201                 if (!empty($links) && is_array($links)) {
 202                         foreach ($links as $link) {
 203                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 204                                         $profile_link = $link['@attributes']['href'];
 205                                 }
 206                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 207                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 208                                 }
 209                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 210                                         $hcard_url = $link['@attributes']['href'];
 211                                 }
 212                         }
 213                 }
 214                 return $profile_link;
 215         }
 216
 217         /**
 218          * @brief Check an URI for LRDD data
 219          *
 220          * this is a replacement for the "lrdd" function.
 221          * It isn't used in this class and has some redundancies in the code.
 222          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 223          *
 224          * @param string $uri Address that should be probed
 225          *
 226          * @return array uri data
 227          * @throws HTTPException\InternalServerErrorException
 228          */
 229         public static function lrdd($uri)
 230         {
 231                 $lrdd = self::hostMeta($uri);
 232                 $webfinger = null;
 233
 234                 if (is_bool($lrdd)) {
 235                         return [];
 236                 }
 237
 238                 if (!$lrdd) {
 239                         $parts = @parse_url($uri);
 240                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 241                                 return [];
 242                         }
 243
 244                         $host = $parts["host"];
 245                         if (!empty($parts["port"])) {
 246                                 $host .= ':'.$parts["port"];
 247                         }
 248
 249                         $path_parts = explode("/", trim($parts["path"], "/"));
 250
 251                         $nick = array_pop($path_parts);
 252
 253                         do {
 254                                 $lrdd = self::hostMeta($host);
 255                                 $host .= "/".array_shift($path_parts);
 256                         } while (!$lrdd && (sizeof($path_parts) > 0));
 257                 }
 258
 259                 if (!$lrdd) {
 260                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
 261                         return [];
 262                 }
 263
 264                 foreach ($lrdd as $type => $template) {
 265                         if ($webfinger) {
 266                                 continue;
 267                         }
 268
 269                         $path = str_replace('{uri}', urlencode($uri), $template);
 270                         $webfinger = self::webfinger($path, $type);
 271
 272                         if (!$webfinger && (strstr($uri, "@"))) {
 273                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 274                                 $webfinger = self::webfinger($path, $type);
 275                         }
 276
 277                         // Special treatment for Mastodon
 278                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 279                         // But the webfinger for this format fails.
 280                         if (!$webfinger && !empty($nick)) {
 281                                 // Mastodon uses a "@" as prefix for usernames in their url format
 282                                 $nick = ltrim($nick, '@');
 283
 284                                 $addr = $nick."@".$host;
 285
 286                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 287                                 $webfinger = self::webfinger($path, $type);
 288                         }
 289                 }
 290
 291                 if (!is_array($webfinger["links"])) {
 292                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
 293                         return false;
 294                 }
 295
 296                 $data = [];
 297
 298                 foreach ($webfinger["links"] as $link) {
 299                         $data[] = ["@attributes" => $link];
 300                 }
 301
 302                 if (is_array($webfinger["aliases"])) {
 303                         foreach ($webfinger["aliases"] as $alias) {
 304                                 $data[] = ["@attributes" =>
 305                                                         ["rel" => "alias",
 306                                                                 "href" => $alias]];
 307                         }
 308                 }
 309
 310                 return $data;
 311         }
 312
 313         /**
 314          * @brief Fetch information (protocol endpoints and user information) about a given uri
 315          *
 316          * @param string  $uri     Address that should be probed
 317          * @param string  $network Test for this specific network
 318          * @param integer $uid     User ID for the probe (only used for mails)
 319          * @param boolean $cache   Use cached values?
 320          *
 321          * @return array uri data
 322          * @throws HTTPException\InternalServerErrorException
 323          * @throws \ImagickException
 324          */
 325         public static function uri($uri, $network = '', $uid = -1, $cache = true)
 326         {
 327                 if ($cache) {
 328                         $result = Cache::get('Probe::uri:' . $network . ':' . $uri);
 329                         if (!is_null($result)) {
 330                                 return $result;
 331                         }
 332                 }
 333
 334                 if ($uid == -1) {
 335                         $uid = local_user();
 336                 }
 337
 338                 self::$istimeout = false;
 339
 340                 if ($network != Protocol::ACTIVITYPUB) {
 341                         $data = self::detect($uri, $network, $uid);
 342                 } else {
 343                         $data = null;
 344                 }
 345
 346                 // When the previous detection process had got a time out
 347                 // we could falsely detect a Friendica profile as AP profile.
 348                 if (!self::$istimeout) {
 349                         $ap_profile = ActivityPub::probeProfile($uri);
 350
 351                         if (!empty($ap_profile) && empty($network) && (defaults($data, 'network', '') != Protocol::DFRN)) {
 352                                 $data = $ap_profile;
 353                         } elseif (!empty($ap_profile)) {
 354                                 $data = array_merge($ap_profile, $data);
 355                         }
 356                 } else {
 357                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 358                 }
 359
 360                 if (!isset($data['url'])) {
 361                         $data['url'] = $uri;
 362                 }
 363
 364                 if (!empty($data['photo'])) {
 365                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink(defaults($data, 'baseurl', '')), Strings::normaliseLink($data['photo']));
 366                 } else {
 367                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 368                 }
 369
 370                 if (empty($data['name'])) {
 371                         if (!empty($data['nick'])) {
 372                                 $data['name'] = $data['nick'];
 373                         }
 374
 375                         if (empty($data['name'])) {
 376                                 $data['name'] = $data['url'];
 377                         }
 378                 }
 379
 380                 if (empty($data['nick'])) {
 381                         $data['nick'] = strtolower($data['name']);
 382
 383                         if (strpos($data['nick'], ' ')) {
 384                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 385                         }
 386                 }
 387
 388                 if (!empty(self::$baseurl)) {
 389                         $data['baseurl'] = self::$baseurl;
 390                 }
 391
 392                 if (empty($data['network'])) {
 393                         $data['network'] = Protocol::PHANTOM;
 394                 }
 395
 396                 $data = self::rearrangeData($data);
 397
 398                 // Only store into the cache if the value seems to be valid
 399                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 400                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 401                 }
 402
 403                 return $data;
 404         }
 405
 406         /**
 407          * @brief Checks if a profile url should be OStatus but only provides partial information
 408          *
 409          * @param array  $webfinger Webfinger data
 410          * @param string $lrdd      Path template for webfinger request
 411          * @param string $type      type
 412          *
 413          * @return array fixed webfinger data
 414          * @throws HTTPException\InternalServerErrorException
 415          */
 416         private static function fixOStatus($webfinger, $lrdd, $type)
 417         {
 418                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 419                         return $webfinger;
 420                 }
 421
 422                 $is_ostatus = false;
 423                 $has_key = false;
 424
 425                 foreach ($webfinger['links'] as $link) {
 426                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 427                                 $is_ostatus = true;
 428                         }
 429                         if ($link['rel'] == 'magic-public-key') {
 430                                 $has_key = true;
 431                         }
 432                 }
 433
 434                 if (!$is_ostatus || $has_key) {
 435                         return $webfinger;
 436                 }
 437
 438                 $url = Network::switchScheme($webfinger['subject']);
 439                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 440                 $webfinger2 = self::webfinger($path, $type);
 441
 442                 // Is the new webfinger detectable as OStatus?
 443                 if (self::ostatus($webfinger2, true)) {
 444                         $webfinger = $webfinger2;
 445                 }
 446
 447                 return $webfinger;
 448         }
 449
 450         /**
 451          * @brief Fetch information (protocol endpoints and user information) about a given uri
 452          *
 453          * This function is only called by the "uri" function that adds caching and rearranging of data.
 454          *
 455          * @param string  $uri     Address that should be probed
 456          * @param string  $network Test for this specific network
 457          * @param integer $uid     User ID for the probe (only used for mails)
 458          *
 459          * @return array uri data
 460          * @throws HTTPException\InternalServerErrorException
 461          */
 462         private static function detect($uri, $network, $uid)
 463         {
 464                 $parts = parse_url($uri);
 465
 466                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 467                         $host = $parts["host"];
 468                         if (!empty($parts["port"])) {
 469                                 $host .= ':'.$parts["port"];
 470                         }
 471
 472                         if ($host == 'twitter.com') {
 473                                 return ["network" => Protocol::TWITTER];
 474                         }
 475                         $lrdd = self::hostMeta($host);
 476
 477                         if (is_bool($lrdd)) {
 478                                 return [];
 479                         }
 480
 481                         $path_parts = explode("/", trim(defaults($parts, 'path', ''), "/"));
 482
 483                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 484                                 $host .= "/".array_shift($path_parts);
 485                                 $lrdd = self::hostMeta($host);
 486                         }
 487                         if (!$lrdd) {
 488                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 489                                 return self::feed($uri);
 490                         }
 491                         $nick = array_pop($path_parts);
 492
 493                         // Mastodon uses a "@" as prefix for usernames in their url format
 494                         $nick = ltrim($nick, '@');
 495
 496                         $addr = $nick."@".$host;
 497                 } elseif (strstr($uri, '@')) {
 498                         // If the URI starts with "mailto:" then jump directly to the mail detection
 499                         if (strpos($uri, 'mailto:') !== false) {
 500                                 $uri = str_replace('mailto:', '', $uri);
 501                                 return self::mail($uri, $uid);
 502                         }
 503
 504                         if ($network == Protocol::MAIL) {
 505                                 return self::mail($uri, $uid);
 506                         }
 507                         // Remove "acct:" from the URI
 508                         $uri = str_replace('acct:', '', $uri);
 509
 510                         $host = substr($uri, strpos($uri, '@') + 1);
 511                         $nick = substr($uri, 0, strpos($uri, '@'));
 512
 513                         if (strpos($uri, '@twitter.com')) {
 514                                 return ["network" => Protocol::TWITTER];
 515                         }
 516                         $lrdd = self::hostMeta($host);
 517
 518                         if (is_bool($lrdd)) {
 519                                 return [];
 520                         }
 521
 522                         if (!$lrdd) {
 523                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 524                                 return self::mail($uri, $uid);
 525                         }
 526                         $addr = $uri;
 527                 } else {
 528                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 529                         return false;
 530                 }
 531
 532                 $webfinger = false;
 533
 534                 /// @todo Do we need the prefix "acct:" or "acct://"?
 535
 536                 foreach ($lrdd as $type => $template) {
 537                         if ($webfinger) {
 538                                 continue;
 539                         }
 540
 541                         // At first try it with the given uri
 542                         $path = str_replace('{uri}', urlencode($uri), $template);
 543                         $webfinger = self::webfinger($path, $type);
 544
 545                         // Fix possible problems with GNU Social probing to wrong scheme
 546                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 547
 548                         // We cannot be sure that the detected address was correct, so we don't use the values
 549                         if ($webfinger && ($uri != $addr)) {
 550                                 $nick = "";
 551                                 $addr = "";
 552                         }
 553
 554                         // Try webfinger with the address (user@domain.tld)
 555                         if (!$webfinger) {
 556                                 $path = str_replace('{uri}', urlencode($addr), $template);
 557                                 $webfinger = self::webfinger($path, $type);
 558                         }
 559
 560                         // Mastodon needs to have it with "acct:"
 561                         if (!$webfinger) {
 562                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 563                                 $webfinger = self::webfinger($path, $type);
 564                         }
 565                 }
 566
 567                 if (!$webfinger) {
 568                         return self::feed($uri);
 569                 }
 570
 571                 $result = false;
 572
 573                 Logger::log("Probing ".$uri, Logger::DEBUG);
 574
 575                 if (in_array($network, ["", Protocol::DFRN])) {
 576                         $result = self::dfrn($webfinger);
 577                 }
 578                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 579                         $result = self::diaspora($webfinger);
 580                 }
 581                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 582                         $result = self::ostatus($webfinger);
 583                 }
 584                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 585                         $result = self::pumpio($webfinger, $addr);
 586                 }
 587                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 588                         $result = self::feed($uri);
 589                 } else {
 590                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 591                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 592                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 593                                 $result["nick"] = $nick;
 594                         }
 595
 596                         if (empty($result["addr"]) && ($addr != "")) {
 597                                 $result["addr"] = $addr;
 598                         }
 599                 }
 600
 601                 if (empty($result["network"])) {
 602                         $result["network"] = Protocol::PHANTOM;
 603                 }
 604
 605                 if (empty($result["url"])) {
 606                         $result["url"] = $uri;
 607                 }
 608
 609                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 610
 611                 if (empty($result["baseurl"])) {
 612                         $pos = strpos($result["url"], $host);
 613                         if ($pos) {
 614                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 615                         }
 616                 }
 617                 return $result;
 618         }
 619
 620         /**
 621          * @brief Perform a webfinger request.
 622          *
 623          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 624          *
 625          * @param string $url  Address that should be probed
 626          * @param string $type type
 627          *
 628          * @return array webfinger data
 629          * @throws HTTPException\InternalServerErrorException
 630          */
 631         private static function webfinger($url, $type)
 632         {
 633                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 634
 635                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 636                 if ($curlResult->isTimeout()) {
 637                         self::$istimeout = true;
 638                         return false;
 639                 }
 640                 $data = $curlResult->getBody();
 641
 642                 $webfinger = json_decode($data, true);
 643                 if (is_array($webfinger)) {
 644                         if (!isset($webfinger["links"])) {
 645                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 646                                 return false;
 647                         }
 648                         return $webfinger;
 649                 }
 650
 651                 // If it is not JSON, maybe it is XML
 652                 $xrd = XML::parseString($data, false);
 653                 if (!is_object($xrd)) {
 654                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 655                         return false;
 656                 }
 657
 658                 $xrd_arr = XML::elementToArray($xrd);
 659                 if (!isset($xrd_arr["xrd"]["link"])) {
 660                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 661                         return false;
 662                 }
 663
 664                 $webfinger = [];
 665
 666                 if (!empty($xrd_arr["xrd"]["subject"])) {
 667                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 668                 }
 669
 670                 if (!empty($xrd_arr["xrd"]["alias"])) {
 671                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 672                 }
 673
 674                 $webfinger["links"] = [];
 675
 676                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 677                         if (!empty($data["@attributes"])) {
 678                                 $attributes = $data["@attributes"];
 679                         } elseif ($value == "@attributes") {
 680                                 $attributes = $data;
 681                         } else {
 682                                 continue;
 683                         }
 684
 685                         $webfinger["links"][] = $attributes;
 686                 }
 687                 return $webfinger;
 688         }
 689
 690         /**
 691          * @brief Poll the Friendica specific noscrape page.
 692          *
 693          * "noscrape" is a faster alternative to fetch the data from the hcard.
 694          * This functionality was originally created for the directory.
 695          *
 696          * @param string $noscrape_url Link to the noscrape page
 697          * @param array  $data         The already fetched data
 698          *
 699          * @return array noscrape data
 700          * @throws HTTPException\InternalServerErrorException
 701          */
 702         private static function pollNoscrape($noscrape_url, $data)
 703         {
 704                 $curlResult = Network::curl($noscrape_url);
 705                 if ($curlResult->isTimeout()) {
 706                         self::$istimeout = true;
 707                         return false;
 708                 }
 709                 $content = $curlResult->getBody();
 710                 if (!$content) {
 711                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 712                         return false;
 713                 }
 714
 715                 $json = json_decode($content, true);
 716                 if (!is_array($json)) {
 717                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 718                         return false;
 719                 }
 720
 721                 if (!empty($json["fn"])) {
 722                         $data["name"] = $json["fn"];
 723                 }
 724
 725                 if (!empty($json["addr"])) {
 726                         $data["addr"] = $json["addr"];
 727                 }
 728
 729                 if (!empty($json["nick"])) {
 730                         $data["nick"] = $json["nick"];
 731                 }
 732
 733                 if (!empty($json["guid"])) {
 734                         $data["guid"] = $json["guid"];
 735                 }
 736
 737                 if (!empty($json["comm"])) {
 738                         $data["community"] = $json["comm"];
 739                 }
 740
 741                 if (!empty($json["tags"])) {
 742                         $keywords = implode(" ", $json["tags"]);
 743                         if ($keywords != "") {
 744                                 $data["keywords"] = $keywords;
 745                         }
 746                 }
 747
 748                 $location = Profile::formatLocation($json);
 749                 if ($location) {
 750                         $data["location"] = $location;
 751                 }
 752
 753                 if (!empty($json["about"])) {
 754                         $data["about"] = $json["about"];
 755                 }
 756
 757                 if (!empty($json["key"])) {
 758                         $data["pubkey"] = $json["key"];
 759                 }
 760
 761                 if (!empty($json["photo"])) {
 762                         $data["photo"] = $json["photo"];
 763                 }
 764
 765                 if (!empty($json["dfrn-request"])) {
 766                         $data["request"] = $json["dfrn-request"];
 767                 }
 768
 769                 if (!empty($json["dfrn-confirm"])) {
 770                         $data["confirm"] = $json["dfrn-confirm"];
 771                 }
 772
 773                 if (!empty($json["dfrn-notify"])) {
 774                         $data["notify"] = $json["dfrn-notify"];
 775                 }
 776
 777                 if (!empty($json["dfrn-poll"])) {
 778                         $data["poll"] = $json["dfrn-poll"];
 779                 }
 780
 781                 return $data;
 782         }
 783
 784         /**
 785          * @brief Check for valid DFRN data
 786          *
 787          * @param array $data DFRN data
 788          *
 789          * @return int Number of errors
 790          */
 791         public static function validDfrn($data)
 792         {
 793                 $errors = 0;
 794                 if (!isset($data['key'])) {
 795                         $errors ++;
 796                 }
 797                 if (!isset($data['dfrn-request'])) {
 798                         $errors ++;
 799                 }
 800                 if (!isset($data['dfrn-confirm'])) {
 801                         $errors ++;
 802                 }
 803                 if (!isset($data['dfrn-notify'])) {
 804                         $errors ++;
 805                 }
 806                 if (!isset($data['dfrn-poll'])) {
 807                         $errors ++;
 808                 }
 809                 return $errors;
 810         }
 811
 812         /**
 813          * @brief Fetch data from a DFRN profile page and via "noscrape"
 814          *
 815          * @param string $profile_link Link to the profile page
 816          *
 817          * @return array profile data
 818          * @throws HTTPException\InternalServerErrorException
 819          * @throws \ImagickException
 820          */
 821         public static function profile($profile_link)
 822         {
 823                 $data = [];
 824
 825                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
 826
 827                 // Fetch data via noscrape - this is faster
 828                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 829                 $data = self::pollNoscrape($noscrape_url, $data);
 830
 831                 if (!isset($data["notify"])
 832                         || !isset($data["confirm"])
 833                         || !isset($data["request"])
 834                         || !isset($data["poll"])
 835                         || !isset($data["name"])
 836                         || !isset($data["photo"])
 837                 ) {
 838                         $data = self::pollHcard($profile_link, $data, true);
 839                 }
 840
 841                 $prof_data = [];
 842
 843                 if (empty($data["addr"]) || empty($data["nick"])) {
 844                         $probe_data = self::uri($profile_link);
 845                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 846                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 847                 }
 848
 849                 $prof_data["addr"]         = $data["addr"];
 850                 $prof_data["nick"]         = $data["nick"];
 851                 $prof_data["dfrn-request"] = defaults($data, 'request', null);
 852                 $prof_data["dfrn-confirm"] = defaults($data, 'confirm', null);
 853                 $prof_data["dfrn-notify"]  = defaults($data, 'notify' , null);
 854                 $prof_data["dfrn-poll"]    = defaults($data, 'poll'   , null);
 855                 $prof_data["photo"]        = defaults($data, 'photo'  , null);
 856                 $prof_data["fn"]           = defaults($data, 'name'   , null);
 857                 $prof_data["key"]          = defaults($data, 'pubkey' , null);
 858
 859                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
 860
 861                 return $prof_data;
 862         }
 863
 864         /**
 865          * @brief Check for DFRN contact
 866          *
 867          * @param array $webfinger Webfinger data
 868          *
 869          * @return array DFRN data
 870          * @throws HTTPException\InternalServerErrorException
 871          */
 872         private static function dfrn($webfinger)
 873         {
 874                 $hcard_url = "";
 875                 $data = [];
 876                 // The array is reversed to take into account the order of preference for same-rel links
 877                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
 878                 foreach (array_reverse($webfinger["links"]) as $link) {
 879                         if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) {
 880                                 $data["network"] = Protocol::DFRN;
 881                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
 882                                 $data["poll"] = $link["href"];
 883                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
 884                                 $data["url"] = $link["href"];
 885                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
 886                                 $hcard_url = $link["href"];
 887                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
 888                                 $data["poco"] = $link["href"];
 889                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
 890                                 $data["photo"] = $link["href"];
 891                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
 892                                 $data["baseurl"] = trim($link["href"], '/');
 893                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
 894                                 $data["guid"] = $link["href"];
 895                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
 896                                 $data["pubkey"] = base64_decode($link["href"]);
 897
 898                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 899                                 if (strstr($data["pubkey"], 'RSA ')) {
 900                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 901                                 }
 902                         }
 903                 }
 904
 905                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 906                         foreach ($webfinger["aliases"] as $alias) {
 907                                 if (empty($data["url"]) && !strstr($alias, "@")) {
 908                                         $data["url"] = $alias;
 909                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
 910                                         $data["alias"] = $alias;
 911                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 912                                         $data["addr"] = substr($alias, 5);
 913                                 }
 914                         }
 915                 }
 916
 917                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 918                         $data["addr"] = substr($webfinger["subject"], 5);
 919                 }
 920
 921                 if (!isset($data["network"]) || ($hcard_url == "")) {
 922                         return false;
 923                 }
 924
 925                 // Fetch data via noscrape - this is faster
 926                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
 927                 $data = self::pollNoscrape($noscrape_url, $data);
 928
 929                 if (isset($data["notify"])
 930                         && isset($data["confirm"])
 931                         && isset($data["request"])
 932                         && isset($data["poll"])
 933                         && isset($data["name"])
 934                         && isset($data["photo"])
 935                 ) {
 936                         return $data;
 937                 }
 938
 939                 $data = self::pollHcard($hcard_url, $data, true);
 940
 941                 return $data;
 942         }
 943
 944         /**
 945          * @brief Poll the hcard page (Diaspora and Friendica specific)
 946          *
 947          * @param string  $hcard_url Link to the hcard page
 948          * @param array   $data      The already fetched data
 949          * @param boolean $dfrn      Poll DFRN specific data
 950          *
 951          * @return array hcard data
 952          * @throws HTTPException\InternalServerErrorException
 953          */
 954         private static function pollHcard($hcard_url, $data, $dfrn = false)
 955         {
 956                 $curlResult = Network::curl($hcard_url);
 957                 if ($curlResult->isTimeout()) {
 958                         self::$istimeout = true;
 959                         return false;
 960                 }
 961                 $content = $curlResult->getBody();
 962                 if (!$content) {
 963                         return false;
 964                 }
 965
 966                 $doc = new DOMDocument();
 967                 if (!@$doc->loadHTML($content)) {
 968                         return false;
 969                 }
 970
 971                 $xpath = new DomXPath($doc);
 972
 973                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 974                 if (!is_object($vcards)) {
 975                         return false;
 976                 }
 977
 978                 if (!isset($data["baseurl"])) {
 979                         $data["baseurl"] = "";
 980                 }
 981
 982                 if ($vcards->length > 0) {
 983                         $vcard = $vcards->item(0);
 984
 985                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 986                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 987                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 988                         if (($search->length > 0) && empty($data["guid"])) {
 989                                 $data["guid"] = $search->item(0)->nodeValue;
 990                         }
 991
 992                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 993                         if ($search->length > 0) {
 994                                 $data["nick"] = $search->item(0)->nodeValue;
 995                         }
 996
 997                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 998                         if ($search->length > 0) {
 999                                 $data["name"] = $search->item(0)->nodeValue;
1000                         }
1001
1002                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1003                         if ($search->length > 0) {
1004                                 $data["searchable"] = $search->item(0)->nodeValue;
1005                         }
1006
1007                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1008                         if ($search->length > 0) {
1009                                 $data["pubkey"] = $search->item(0)->nodeValue;
1010                                 if (strstr($data["pubkey"], 'RSA ')) {
1011                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1012                                 }
1013                         }
1014
1015                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1016                         if ($search->length > 0) {
1017                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1018                         }
1019                 }
1020
1021                 $avatar = [];
1022                 if (!empty($vcard)) {
1023                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1024                         foreach ($photos as $photo) {
1025                                 $attr = [];
1026                                 foreach ($photo->attributes as $attribute) {
1027                                         $attr[$attribute->name] = trim($attribute->value);
1028                                 }
1029
1030                                 if (isset($attr["src"]) && isset($attr["width"])) {
1031                                         $avatar[$attr["width"]] = $attr["src"];
1032                                 }
1033
1034                                 // We don't have a width. So we just take everything that we got.
1035                                 // This is a Hubzilla workaround which doesn't send a width.
1036                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1037                                         $avatar[] = $attr["src"];
1038                                 }
1039                         }
1040                 }
1041
1042                 if (sizeof($avatar)) {
1043                         ksort($avatar);
1044                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1045                 }
1046
1047                 if ($dfrn) {
1048                         // Poll DFRN specific data
1049                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1050                         if ($search->length > 0) {
1051                                 foreach ($search as $link) {
1052                                         //$data["request"] = $search->item(0)->nodeValue;
1053                                         $attr = [];
1054                                         foreach ($link->attributes as $attribute) {
1055                                                 $attr[$attribute->name] = trim($attribute->value);
1056                                         }
1057
1058                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1059                                 }
1060                         }
1061
1062                         // Older Friendica versions had used the "uid" field differently than newer versions
1063                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1064                                 unset($data["guid"]);
1065                         }
1066                 }
1067
1068
1069                 return $data;
1070         }
1071
1072         /**
1073          * @brief Check for Diaspora contact
1074          *
1075          * @param array $webfinger Webfinger data
1076          *
1077          * @return array Diaspora data
1078          * @throws HTTPException\InternalServerErrorException
1079          */
1080         private static function diaspora($webfinger)
1081         {
1082                 $hcard_url = "";
1083                 $data = [];
1084                 // The array is reversed to take into account the order of preference for same-rel links
1085                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1086                 foreach (array_reverse($webfinger["links"]) as $link) {
1087                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1088                                 $hcard_url = $link["href"];
1089                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1090                                 $data["baseurl"] = trim($link["href"], '/');
1091                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1092                                 $data["guid"] = $link["href"];
1093                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1094                                 $data["url"] = $link["href"];
1095                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1096                                 $data["poll"] = $link["href"];
1097                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1098                                 $data["poco"] = $link["href"];
1099                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1100                                 $data["notify"] = $link["href"];
1101                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1102                                 $data["pubkey"] = base64_decode($link["href"]);
1103
1104                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1105                                 if (strstr($data["pubkey"], 'RSA ')) {
1106                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1107                                 }
1108                         }
1109                 }
1110
1111                 if (!isset($data["url"]) || ($hcard_url == "")) {
1112                         return false;
1113                 }
1114
1115                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1116                         foreach ($webfinger["aliases"] as $alias) {
1117                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1118                                         $data["alias"] = $alias;
1119                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1120                                         $data["addr"] = substr($alias, 5);
1121                                 }
1122                         }
1123                 }
1124
1125                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1126                         $data["addr"] = substr($webfinger["subject"], 5);
1127                 }
1128
1129                 // Fetch further information from the hcard
1130                 $data = self::pollHcard($hcard_url, $data);
1131
1132                 if (!$data) {
1133                         return false;
1134                 }
1135
1136                 if (isset($data["url"])
1137                         && isset($data["guid"])
1138                         && isset($data["baseurl"])
1139                         && isset($data["pubkey"])
1140                         && ($hcard_url != "")
1141                 ) {
1142                         $data["network"] = Protocol::DIASPORA;
1143
1144                         // The Diaspora handle must always be lowercase
1145                         if (!empty($data["addr"])) {
1146                                 $data["addr"] = strtolower($data["addr"]);
1147                         }
1148
1149                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1150                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1151                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1152                 } else {
1153                         return false;
1154                 }
1155
1156                 return $data;
1157         }
1158
1159         /**
1160          * @brief Check for OStatus contact
1161          *
1162          * @param array $webfinger Webfinger data
1163          * @param bool  $short     Short detection mode
1164          *
1165          * @return array|bool OStatus data or "false" on error or "true" on short mode
1166          * @throws HTTPException\InternalServerErrorException
1167          */
1168         private static function ostatus($webfinger, $short = false)
1169         {
1170                 $data = [];
1171
1172                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1173                         foreach ($webfinger["aliases"] as $alias) {
1174                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1175                                         $data["addr"] = str_replace('acct:', '', $alias);
1176                                 }
1177                         }
1178                 }
1179
1180                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1181                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1182                 ) {
1183                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1184                 }
1185
1186                 if (is_array($webfinger["links"])) {
1187                         // The array is reversed to take into account the order of preference for same-rel links
1188                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1189                         foreach (array_reverse($webfinger["links"]) as $link) {
1190                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1191                                         && (defaults($link, "type", "") == "text/html")
1192                                         && ($link["href"] != "")
1193                                 ) {
1194                                         $data["url"] = $link["href"];
1195                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1196                                         $data["notify"] = $link["href"];
1197                                 } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1198                                         $data["poll"] = $link["href"];
1199                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1200                                         $pubkey = $link["href"];
1201
1202                                         if (substr($pubkey, 0, 5) === 'data:') {
1203                                                 if (strstr($pubkey, ',')) {
1204                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1205                                                 } else {
1206                                                         $pubkey = substr($pubkey, 5);
1207                                                 }
1208                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1209                                                 $curlResult = Network::curl($pubkey);
1210                                                 if ($curlResult->isTimeout()) {
1211                                                         self::$istimeout = true;
1212                                                         return false;
1213                                                 }
1214                                                 $pubkey = $curlResult->getBody();
1215                                         }
1216
1217                                         $key = explode(".", $pubkey);
1218
1219                                         if (sizeof($key) >= 3) {
1220                                                 $m = Strings::base64UrlDecode($key[1]);
1221                                                 $e = Strings::base64UrlDecode($key[2]);
1222                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1223                                         }
1224                                 }
1225                         }
1226                 }
1227
1228                 if (isset($data["notify"]) && isset($data["pubkey"])
1229                         && isset($data["poll"])
1230                         && isset($data["url"])
1231                 ) {
1232                         $data["network"] = Protocol::OSTATUS;
1233                 } else {
1234                         return false;
1235                 }
1236
1237                 if ($short) {
1238                         return true;
1239                 }
1240
1241                 // Fetch all additional data from the feed
1242                 $curlResult = Network::curl($data["poll"]);
1243                 if ($curlResult->isTimeout()) {
1244                         self::$istimeout = true;
1245                         return false;
1246                 }
1247                 $feed = $curlResult->getBody();
1248                 $dummy1 = null;
1249                 $dummy2 = null;
1250                 $dummy2 = null;
1251                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1252                 if (!$feed_data) {
1253                         return false;
1254                 }
1255
1256                 if (!empty($feed_data["header"]["author-name"])) {
1257                         $data["name"] = $feed_data["header"]["author-name"];
1258                 }
1259                 if (!empty($feed_data["header"]["author-nick"])) {
1260                         $data["nick"] = $feed_data["header"]["author-nick"];
1261                 }
1262                 if (!empty($feed_data["header"]["author-avatar"])) {
1263                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1264                 }
1265                 if (!empty($feed_data["header"]["author-id"])) {
1266                         $data["alias"] = $feed_data["header"]["author-id"];
1267                 }
1268                 if (!empty($feed_data["header"]["author-location"])) {
1269                         $data["location"] = $feed_data["header"]["author-location"];
1270                 }
1271                 if (!empty($feed_data["header"]["author-about"])) {
1272                         $data["about"] = $feed_data["header"]["author-about"];
1273                 }
1274                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1275                 // So we take the value that we just fetched, although the other one worked as well
1276                 if (!empty($feed_data["header"]["author-link"])) {
1277                         $data["url"] = $feed_data["header"]["author-link"];
1278                 }
1279
1280                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1281                         $data['url'] = $data["alias"];
1282                         $data["alias"] = '';
1283                 }
1284
1285                 /// @todo Fetch location and "about" from the feed as well
1286                 return $data;
1287         }
1288
1289         /**
1290          * @brief Fetch data from a pump.io profile page
1291          *
1292          * @param string $profile_link Link to the profile page
1293          *
1294          * @return array profile data
1295          */
1296         private static function pumpioProfileData($profile_link)
1297         {
1298                 $doc = new DOMDocument();
1299                 if (!@$doc->loadHTMLFile($profile_link)) {
1300                         return false;
1301                 }
1302
1303                 $xpath = new DomXPath($doc);
1304
1305                 $data = [];
1306
1307                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1308
1309                 if ($data["name"] == '') {
1310                         // This is ugly - but pump.io doesn't seem to know a better way for it
1311                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1312                         $pos = strpos($data["name"], chr(10));
1313                         if ($pos) {
1314                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1315                         }
1316                 }
1317
1318                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1319
1320                 if ($data["location"] == '') {
1321                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1322                 }
1323
1324                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1325
1326                 if ($data["about"] == '') {
1327                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1328                 }
1329
1330                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1331                 if (!$avatar) {
1332                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1333                 }
1334                 if ($avatar) {
1335                         foreach ($avatar->attributes as $attribute) {
1336                                 if ($attribute->name == "src") {
1337                                         $data["photo"] = trim($attribute->value);
1338                                 }
1339                         }
1340                 }
1341
1342                 return $data;
1343         }
1344
1345         /**
1346          * @brief Check for pump.io contact
1347          *
1348          * @param array $webfinger Webfinger data
1349          *
1350          * @param       $addr
1351          * @return array pump.io data
1352          */
1353         private static function pumpio($webfinger, $addr)
1354         {
1355                 $data = [];
1356                 // The array is reversed to take into account the order of preference for same-rel links
1357                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1358                 foreach (array_reverse($webfinger["links"]) as $link) {
1359                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1360                                 && (defaults($link, "type", "") == "text/html")
1361                                 && ($link["href"] != "")
1362                         ) {
1363                                 $data["url"] = $link["href"];
1364                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1365                                 $data["notify"] = $link["href"];
1366                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1367                                 $data["poll"] = $link["href"];
1368                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1369                                 $data["dialback"] = $link["href"];
1370                         }
1371                 }
1372                 if (isset($data["poll"]) && isset($data["notify"])
1373                         && isset($data["dialback"])
1374                         && isset($data["url"])
1375                 ) {
1376                         // by now we use these fields only for the network type detection
1377                         // So we unset all data that isn't used at the moment
1378                         unset($data["dialback"]);
1379
1380                         $data["network"] = Protocol::PUMPIO;
1381                 } else {
1382                         return false;
1383                 }
1384
1385                 $profile_data = self::pumpioProfileData($data["url"]);
1386
1387                 if (!$profile_data) {
1388                         return false;
1389                 }
1390
1391                 $data = array_merge($data, $profile_data);
1392
1393                 if (($addr != '') && ($data['name'] != '')) {
1394                         $name = trim(str_replace($addr, '', $data['name']));
1395                         if ($name != '') {
1396                                 $data['name'] = $name;
1397                         }
1398                 }
1399
1400                 return $data;
1401         }
1402
1403         /**
1404          * @brief Check page for feed link
1405          *
1406          * @param string $url Page link
1407          *
1408          * @return string feed link
1409          */
1410         private static function getFeedLink($url)
1411         {
1412                 $doc = new DOMDocument();
1413
1414                 if (!@$doc->loadHTMLFile($url)) {
1415                         return false;
1416                 }
1417
1418                 $xpath = new DomXPath($doc);
1419
1420                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1421                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1422                 if (!is_object($feeds)) {
1423                         return false;
1424                 }
1425
1426                 if ($feeds->length == 0) {
1427                         return false;
1428                 }
1429
1430                 $feed_url = "";
1431
1432                 foreach ($feeds as $feed) {
1433                         $attr = [];
1434                         foreach ($feed->attributes as $attribute) {
1435                                 $attr[$attribute->name] = trim($attribute->value);
1436                         }
1437
1438                         if ($feed_url == "") {
1439                                 $feed_url = $attr["href"];
1440                         }
1441                 }
1442
1443                 return $feed_url;
1444         }
1445
1446         /**
1447          * @brief Check for feed contact
1448          *
1449          * @param string  $url   Profile link
1450          * @param boolean $probe Do a probe if the page contains a feed link
1451          *
1452          * @return array feed data
1453          * @throws HTTPException\InternalServerErrorException
1454          */
1455         private static function feed($url, $probe = true)
1456         {
1457                 $curlResult = Network::curl($url);
1458                 if ($curlResult->isTimeout()) {
1459                         self::$istimeout = true;
1460                         return false;
1461                 }
1462                 $feed = $curlResult->getBody();
1463                 $dummy1 = $dummy2 = $dummy3 = null;
1464                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1465
1466                 if (!$feed_data) {
1467                         if (!$probe) {
1468                                 return false;
1469                         }
1470
1471                         $feed_url = self::getFeedLink($url);
1472
1473                         if (!$feed_url) {
1474                                 return false;
1475                         }
1476
1477                         return self::feed($feed_url, false);
1478                 }
1479
1480                 if (!empty($feed_data["header"]["author-name"])) {
1481                         $data["name"] = $feed_data["header"]["author-name"];
1482                 }
1483
1484                 if (!empty($feed_data["header"]["author-nick"])) {
1485                         $data["nick"] = $feed_data["header"]["author-nick"];
1486                 }
1487
1488                 if (!empty($feed_data["header"]["author-avatar"])) {
1489                         $data["photo"] = $feed_data["header"]["author-avatar"];
1490                 }
1491
1492                 if (!empty($feed_data["header"]["author-id"])) {
1493                         $data["alias"] = $feed_data["header"]["author-id"];
1494                 }
1495
1496                 $data["url"] = $url;
1497                 $data["poll"] = $url;
1498
1499                 if (!empty($feed_data["header"]["author-link"])) {
1500                         $data["baseurl"] = $feed_data["header"]["author-link"];
1501                 } else {
1502                         $data["baseurl"] = $data["url"];
1503                 }
1504
1505                 $data["network"] = Protocol::FEED;
1506
1507                 return $data;
1508         }
1509
1510         /**
1511          * @brief Check for mail contact
1512          *
1513          * @param string  $uri Profile link
1514          * @param integer $uid User ID
1515          *
1516          * @return array mail data
1517          * @throws \Exception
1518          */
1519         private static function mail($uri, $uid)
1520         {
1521                 if (!Network::isEmailDomainValid($uri)) {
1522                         return false;
1523                 }
1524
1525                 if ($uid == 0) {
1526                         return false;
1527                 }
1528
1529                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1530
1531                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1532                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1533                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1534
1535                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1536                         return false;
1537                 }
1538
1539                 $mailbox = Email::constructMailboxName($mailacct);
1540                 $password = '';
1541                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1542                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1543                 if (!$mbox) {
1544                         return false;
1545                 }
1546
1547                 $msgs = Email::poll($mbox, $uri);
1548                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1549
1550                 if (!count($msgs)) {
1551                         return false;
1552                 }
1553
1554                 $phost = substr($uri, strpos($uri, '@') + 1);
1555
1556                 $data = [];
1557                 $data["addr"]    = $uri;
1558                 $data["network"] = Protocol::MAIL;
1559                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1560                 $data["nick"]    = $data["name"];
1561                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1562                 $data["url"]     = 'mailto:'.$uri;
1563                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1564                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1565
1566                 $x = Email::messageMeta($mbox, $msgs[0]);
1567                 if (stristr($x[0]->from, $uri)) {
1568                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1569                 } elseif (stristr($x[0]->to, $uri)) {
1570                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1571                 }
1572                 if (isset($adr)) {
1573                         foreach ($adr as $feadr) {
1574                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1575                                         &&(strcasecmp($feadr->host, $phost) == 0)
1576                                         && (strlen($feadr->personal))
1577                                 ) {
1578                                         $personal = imap_mime_header_decode($feadr->personal);
1579                                         $data["name"] = "";
1580                                         foreach ($personal as $perspart) {
1581                                                 if ($perspart->charset != "default") {
1582                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1583                                                 } else {
1584                                                         $data["name"] .= $perspart->text;
1585                                                 }
1586                                         }
1587
1588                                         $data["name"] = Strings::escapeTags($data["name"]);
1589                                 }
1590                         }
1591                 }
1592                 if (!empty($mbox)) {
1593                         imap_close($mbox);
1594                 }
1595                 return $data;
1596         }
1597
1598         /**
1599          * @brief Mix two paths together to possibly fix missing parts
1600          *
1601          * @param string $avatar Path to the avatar
1602          * @param string $base   Another path that is hopefully complete
1603          *
1604          * @return string fixed avatar path
1605          * @throws \Exception
1606          */
1607         public static function fixAvatar($avatar, $base)
1608         {
1609                 $base_parts = parse_url($base);
1610
1611                 // Remove all parts that could create a problem
1612                 unset($base_parts['path']);
1613                 unset($base_parts['query']);
1614                 unset($base_parts['fragment']);
1615
1616                 $avatar_parts = parse_url($avatar);
1617
1618                 // Now we mix them
1619                 $parts = array_merge($base_parts, $avatar_parts);
1620
1621                 // And put them together again
1622                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1623                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1624                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1625                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1626                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1627                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1628
1629                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1630
1631                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1632
1633                 return $fixed;
1634         }
1635 }