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", "account-type",
  49                                 "community", "keywords", "location", "about", "gender", "hide",
  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                                 $ap_profile['batch'] = '';
 355                                 $data = array_merge($ap_profile, $data);
 356                         }
 357                 } else {
 358                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 359                 }
 360
 361                 if (!isset($data['url'])) {
 362                         $data['url'] = $uri;
 363                 }
 364
 365                 if (!empty($data['photo'])) {
 366                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink(defaults($data, 'baseurl', '')), Strings::normaliseLink($data['photo']));
 367                 } else {
 368                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 369                 }
 370
 371                 if (empty($data['name'])) {
 372                         if (!empty($data['nick'])) {
 373                                 $data['name'] = $data['nick'];
 374                         }
 375
 376                         if (empty($data['name'])) {
 377                                 $data['name'] = $data['url'];
 378                         }
 379                 }
 380
 381                 if (empty($data['nick'])) {
 382                         $data['nick'] = strtolower($data['name']);
 383
 384                         if (strpos($data['nick'], ' ')) {
 385                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 386                         }
 387                 }
 388
 389                 if (!empty(self::$baseurl)) {
 390                         $data['baseurl'] = self::$baseurl;
 391                 }
 392
 393                 if (empty($data['network'])) {
 394                         $data['network'] = Protocol::PHANTOM;
 395                 }
 396
 397                 $data = self::rearrangeData($data);
 398
 399                 // Only store into the cache if the value seems to be valid
 400                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 401                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 402                 }
 403
 404                 return $data;
 405         }
 406
 407         /**
 408          * @brief Checks if a profile url should be OStatus but only provides partial information
 409          *
 410          * @param array  $webfinger Webfinger data
 411          * @param string $lrdd      Path template for webfinger request
 412          * @param string $type      type
 413          *
 414          * @return array fixed webfinger data
 415          * @throws HTTPException\InternalServerErrorException
 416          */
 417         private static function fixOStatus($webfinger, $lrdd, $type)
 418         {
 419                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 420                         return $webfinger;
 421                 }
 422
 423                 $is_ostatus = false;
 424                 $has_key = false;
 425
 426                 foreach ($webfinger['links'] as $link) {
 427                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 428                                 $is_ostatus = true;
 429                         }
 430                         if ($link['rel'] == 'magic-public-key') {
 431                                 $has_key = true;
 432                         }
 433                 }
 434
 435                 if (!$is_ostatus || $has_key) {
 436                         return $webfinger;
 437                 }
 438
 439                 $url = Network::switchScheme($webfinger['subject']);
 440                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 441                 $webfinger2 = self::webfinger($path, $type);
 442
 443                 // Is the new webfinger detectable as OStatus?
 444                 if (self::ostatus($webfinger2, true)) {
 445                         $webfinger = $webfinger2;
 446                 }
 447
 448                 return $webfinger;
 449         }
 450
 451         /**
 452          * @brief Fetch information (protocol endpoints and user information) about a given uri
 453          *
 454          * This function is only called by the "uri" function that adds caching and rearranging of data.
 455          *
 456          * @param string  $uri     Address that should be probed
 457          * @param string  $network Test for this specific network
 458          * @param integer $uid     User ID for the probe (only used for mails)
 459          *
 460          * @return array uri data
 461          * @throws HTTPException\InternalServerErrorException
 462          */
 463         private static function detect($uri, $network, $uid)
 464         {
 465                 $parts = parse_url($uri);
 466
 467                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 468                         $host = $parts["host"];
 469                         if (!empty($parts["port"])) {
 470                                 $host .= ':'.$parts["port"];
 471                         }
 472
 473                         if ($host == 'twitter.com') {
 474                                 return ["network" => Protocol::TWITTER];
 475                         }
 476                         $lrdd = self::hostMeta($host);
 477
 478                         if (is_bool($lrdd)) {
 479                                 return [];
 480                         }
 481
 482                         $path_parts = explode("/", trim(defaults($parts, 'path', ''), "/"));
 483
 484                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 485                                 $host .= "/".array_shift($path_parts);
 486                                 $lrdd = self::hostMeta($host);
 487                         }
 488                         if (!$lrdd) {
 489                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 490                                 return self::feed($uri);
 491                         }
 492                         $nick = array_pop($path_parts);
 493
 494                         // Mastodon uses a "@" as prefix for usernames in their url format
 495                         $nick = ltrim($nick, '@');
 496
 497                         $addr = $nick."@".$host;
 498                 } elseif (strstr($uri, '@')) {
 499                         // If the URI starts with "mailto:" then jump directly to the mail detection
 500                         if (strpos($uri, 'mailto:') !== false) {
 501                                 $uri = str_replace('mailto:', '', $uri);
 502                                 return self::mail($uri, $uid);
 503                         }
 504
 505                         if ($network == Protocol::MAIL) {
 506                                 return self::mail($uri, $uid);
 507                         }
 508                         // Remove "acct:" from the URI
 509                         $uri = str_replace('acct:', '', $uri);
 510
 511                         $host = substr($uri, strpos($uri, '@') + 1);
 512                         $nick = substr($uri, 0, strpos($uri, '@'));
 513
 514                         if (strpos($uri, '@twitter.com')) {
 515                                 return ["network" => Protocol::TWITTER];
 516                         }
 517                         $lrdd = self::hostMeta($host);
 518
 519                         if (is_bool($lrdd)) {
 520                                 return [];
 521                         }
 522
 523                         if (!$lrdd) {
 524                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 525                                 return self::mail($uri, $uid);
 526                         }
 527                         $addr = $uri;
 528                 } else {
 529                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 530                         return false;
 531                 }
 532
 533                 $webfinger = false;
 534
 535                 /// @todo Do we need the prefix "acct:" or "acct://"?
 536
 537                 foreach ($lrdd as $type => $template) {
 538                         if ($webfinger) {
 539                                 continue;
 540                         }
 541
 542                         // At first try it with the given uri
 543                         $path = str_replace('{uri}', urlencode($uri), $template);
 544                         $webfinger = self::webfinger($path, $type);
 545
 546                         // Fix possible problems with GNU Social probing to wrong scheme
 547                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 548
 549                         // We cannot be sure that the detected address was correct, so we don't use the values
 550                         if ($webfinger && ($uri != $addr)) {
 551                                 $nick = "";
 552                                 $addr = "";
 553                         }
 554
 555                         // Try webfinger with the address (user@domain.tld)
 556                         if (!$webfinger) {
 557                                 $path = str_replace('{uri}', urlencode($addr), $template);
 558                                 $webfinger = self::webfinger($path, $type);
 559                         }
 560
 561                         // Mastodon needs to have it with "acct:"
 562                         if (!$webfinger) {
 563                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 564                                 $webfinger = self::webfinger($path, $type);
 565                         }
 566                 }
 567
 568                 if (!$webfinger) {
 569                         return self::feed($uri);
 570                 }
 571
 572                 $result = false;
 573
 574                 Logger::log("Probing ".$uri, Logger::DEBUG);
 575
 576                 if (in_array($network, ["", Protocol::DFRN])) {
 577                         $result = self::dfrn($webfinger);
 578                 }
 579                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 580                         $result = self::diaspora($webfinger);
 581                 }
 582                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 583                         $result = self::ostatus($webfinger);
 584                 }
 585                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 586                         $result = self::pumpio($webfinger, $addr);
 587                 }
 588                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 589                         $result = self::feed($uri);
 590                 } else {
 591                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 592                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 593                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 594                                 $result["nick"] = $nick;
 595                         }
 596
 597                         if (empty($result["addr"]) && ($addr != "")) {
 598                                 $result["addr"] = $addr;
 599                         }
 600                 }
 601
 602                 if (empty($result["network"])) {
 603                         $result["network"] = Protocol::PHANTOM;
 604                 }
 605
 606                 if (empty($result["url"])) {
 607                         $result["url"] = $uri;
 608                 }
 609
 610                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 611
 612                 if (empty($result["baseurl"])) {
 613                         $pos = strpos($result["url"], $host);
 614                         if ($pos) {
 615                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 616                         }
 617                 }
 618                 return $result;
 619         }
 620
 621         /**
 622          * @brief Perform a webfinger request.
 623          *
 624          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 625          *
 626          * @param string $url  Address that should be probed
 627          * @param string $type type
 628          *
 629          * @return array webfinger data
 630          * @throws HTTPException\InternalServerErrorException
 631          */
 632         private static function webfinger($url, $type)
 633         {
 634                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 635
 636                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 637                 if ($curlResult->isTimeout()) {
 638                         self::$istimeout = true;
 639                         return false;
 640                 }
 641                 $data = $curlResult->getBody();
 642
 643                 $webfinger = json_decode($data, true);
 644                 if (is_array($webfinger)) {
 645                         if (!isset($webfinger["links"])) {
 646                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 647                                 return false;
 648                         }
 649                         return $webfinger;
 650                 }
 651
 652                 // If it is not JSON, maybe it is XML
 653                 $xrd = XML::parseString($data, false);
 654                 if (!is_object($xrd)) {
 655                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 656                         return false;
 657                 }
 658
 659                 $xrd_arr = XML::elementToArray($xrd);
 660                 if (!isset($xrd_arr["xrd"]["link"])) {
 661                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 662                         return false;
 663                 }
 664
 665                 $webfinger = [];
 666
 667                 if (!empty($xrd_arr["xrd"]["subject"])) {
 668                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 669                 }
 670
 671                 if (!empty($xrd_arr["xrd"]["alias"])) {
 672                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 673                 }
 674
 675                 $webfinger["links"] = [];
 676
 677                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 678                         if (!empty($data["@attributes"])) {
 679                                 $attributes = $data["@attributes"];
 680                         } elseif ($value == "@attributes") {
 681                                 $attributes = $data;
 682                         } else {
 683                                 continue;
 684                         }
 685
 686                         $webfinger["links"][] = $attributes;
 687                 }
 688                 return $webfinger;
 689         }
 690
 691         /**
 692          * @brief Poll the Friendica specific noscrape page.
 693          *
 694          * "noscrape" is a faster alternative to fetch the data from the hcard.
 695          * This functionality was originally created for the directory.
 696          *
 697          * @param string $noscrape_url Link to the noscrape page
 698          * @param array  $data         The already fetched data
 699          *
 700          * @return array noscrape data
 701          * @throws HTTPException\InternalServerErrorException
 702          */
 703         private static function pollNoscrape($noscrape_url, $data)
 704         {
 705                 $curlResult = Network::curl($noscrape_url);
 706                 if ($curlResult->isTimeout()) {
 707                         self::$istimeout = true;
 708                         return false;
 709                 }
 710                 $content = $curlResult->getBody();
 711                 if (!$content) {
 712                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 713                         return false;
 714                 }
 715
 716                 $json = json_decode($content, true);
 717                 if (!is_array($json)) {
 718                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 719                         return false;
 720                 }
 721
 722                 if (!empty($json["fn"])) {
 723                         $data["name"] = $json["fn"];
 724                 }
 725
 726                 if (!empty($json["addr"])) {
 727                         $data["addr"] = $json["addr"];
 728                 }
 729
 730                 if (!empty($json["nick"])) {
 731                         $data["nick"] = $json["nick"];
 732                 }
 733
 734                 if (!empty($json["guid"])) {
 735                         $data["guid"] = $json["guid"];
 736                 }
 737
 738                 if (!empty($json["comm"])) {
 739                         $data["community"] = $json["comm"];
 740                 }
 741
 742                 if (!empty($json["tags"])) {
 743                         $keywords = implode(", ", $json["tags"]);
 744                         if ($keywords != "") {
 745                                 $data["keywords"] = $keywords;
 746                         }
 747                 }
 748
 749                 $location = Profile::formatLocation($json);
 750                 if ($location) {
 751                         $data["location"] = $location;
 752                 }
 753
 754                 if (!empty($json["about"])) {
 755                         $data["about"] = $json["about"];
 756                 }
 757
 758                 if (!empty($json["gender"])) {
 759                         $data["gender"] = $json["gender"];
 760                 }
 761
 762                 if (!empty($json["key"])) {
 763                         $data["pubkey"] = $json["key"];
 764                 }
 765
 766                 if (!empty($json["photo"])) {
 767                         $data["photo"] = $json["photo"];
 768                 }
 769
 770                 if (!empty($json["dfrn-request"])) {
 771                         $data["request"] = $json["dfrn-request"];
 772                 }
 773
 774                 if (!empty($json["dfrn-confirm"])) {
 775                         $data["confirm"] = $json["dfrn-confirm"];
 776                 }
 777
 778                 if (!empty($json["dfrn-notify"])) {
 779                         $data["notify"] = $json["dfrn-notify"];
 780                 }
 781
 782                 if (!empty($json["dfrn-poll"])) {
 783                         $data["poll"] = $json["dfrn-poll"];
 784                 }
 785
 786                 if (isset($json["hide"])) {
 787                         $data["hide"] = (bool)$json["hide"];
 788                 } else {
 789                         $data["hide"] = false;
 790                 }
 791
 792                 return $data;
 793         }
 794
 795         /**
 796          * @brief Check for valid DFRN data
 797          *
 798          * @param array $data DFRN data
 799          *
 800          * @return int Number of errors
 801          */
 802         public static function validDfrn($data)
 803         {
 804                 $errors = 0;
 805                 if (!isset($data['key'])) {
 806                         $errors ++;
 807                 }
 808                 if (!isset($data['dfrn-request'])) {
 809                         $errors ++;
 810                 }
 811                 if (!isset($data['dfrn-confirm'])) {
 812                         $errors ++;
 813                 }
 814                 if (!isset($data['dfrn-notify'])) {
 815                         $errors ++;
 816                 }
 817                 if (!isset($data['dfrn-poll'])) {
 818                         $errors ++;
 819                 }
 820                 return $errors;
 821         }
 822
 823         /**
 824          * @brief Fetch data from a DFRN profile page and via "noscrape"
 825          *
 826          * @param string $profile_link Link to the profile page
 827          *
 828          * @return array profile data
 829          * @throws HTTPException\InternalServerErrorException
 830          * @throws \ImagickException
 831          */
 832         public static function profile($profile_link)
 833         {
 834                 $data = [];
 835
 836                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
 837
 838                 // Fetch data via noscrape - this is faster
 839                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 840                 $data = self::pollNoscrape($noscrape_url, $data);
 841
 842                 if (!isset($data["notify"])
 843                         || !isset($data["confirm"])
 844                         || !isset($data["request"])
 845                         || !isset($data["poll"])
 846                         || !isset($data["name"])
 847                         || !isset($data["photo"])
 848                 ) {
 849                         $data = self::pollHcard($profile_link, $data, true);
 850                 }
 851
 852                 $prof_data = [];
 853
 854                 if (empty($data["addr"]) || empty($data["nick"])) {
 855                         $probe_data = self::uri($profile_link);
 856                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 857                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 858                 }
 859
 860                 $prof_data["addr"]         = $data["addr"];
 861                 $prof_data["nick"]         = $data["nick"];
 862                 $prof_data["dfrn-request"] = defaults($data, 'request', null);
 863                 $prof_data["dfrn-confirm"] = defaults($data, 'confirm', null);
 864                 $prof_data["dfrn-notify"]  = defaults($data, 'notify' , null);
 865                 $prof_data["dfrn-poll"]    = defaults($data, 'poll'   , null);
 866                 $prof_data["photo"]        = defaults($data, 'photo'  , null);
 867                 $prof_data["fn"]           = defaults($data, 'name'   , null);
 868                 $prof_data["key"]          = defaults($data, 'pubkey' , null);
 869
 870                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
 871
 872                 return $prof_data;
 873         }
 874
 875         /**
 876          * @brief Check for DFRN contact
 877          *
 878          * @param array $webfinger Webfinger data
 879          *
 880          * @return array DFRN data
 881          * @throws HTTPException\InternalServerErrorException
 882          */
 883         private static function dfrn($webfinger)
 884         {
 885                 $hcard_url = "";
 886                 $data = [];
 887                 // The array is reversed to take into account the order of preference for same-rel links
 888                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
 889                 foreach (array_reverse($webfinger["links"]) as $link) {
 890                         if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) {
 891                                 $data["network"] = Protocol::DFRN;
 892                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
 893                                 $data["poll"] = $link["href"];
 894                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
 895                                 $data["url"] = $link["href"];
 896                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
 897                                 $hcard_url = $link["href"];
 898                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
 899                                 $data["poco"] = $link["href"];
 900                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
 901                                 $data["photo"] = $link["href"];
 902                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
 903                                 $data["baseurl"] = trim($link["href"], '/');
 904                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
 905                                 $data["guid"] = $link["href"];
 906                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
 907                                 $data["pubkey"] = base64_decode($link["href"]);
 908
 909                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 910                                 if (strstr($data["pubkey"], 'RSA ')) {
 911                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 912                                 }
 913                         }
 914                 }
 915
 916                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 917                         foreach ($webfinger["aliases"] as $alias) {
 918                                 if (empty($data["url"]) && !strstr($alias, "@")) {
 919                                         $data["url"] = $alias;
 920                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
 921                                         $data["alias"] = $alias;
 922                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 923                                         $data["addr"] = substr($alias, 5);
 924                                 }
 925                         }
 926                 }
 927
 928                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 929                         $data["addr"] = substr($webfinger["subject"], 5);
 930                 }
 931
 932                 if (!isset($data["network"]) || ($hcard_url == "")) {
 933                         return false;
 934                 }
 935
 936                 // Fetch data via noscrape - this is faster
 937                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
 938                 $data = self::pollNoscrape($noscrape_url, $data);
 939
 940                 if (isset($data["notify"])
 941                         && isset($data["confirm"])
 942                         && isset($data["request"])
 943                         && isset($data["poll"])
 944                         && isset($data["name"])
 945                         && isset($data["photo"])
 946                 ) {
 947                         return $data;
 948                 }
 949
 950                 $data = self::pollHcard($hcard_url, $data, true);
 951
 952                 return $data;
 953         }
 954
 955         /**
 956          * @brief Poll the hcard page (Diaspora and Friendica specific)
 957          *
 958          * @param string  $hcard_url Link to the hcard page
 959          * @param array   $data      The already fetched data
 960          * @param boolean $dfrn      Poll DFRN specific data
 961          *
 962          * @return array hcard data
 963          * @throws HTTPException\InternalServerErrorException
 964          */
 965         private static function pollHcard($hcard_url, $data, $dfrn = false)
 966         {
 967                 $curlResult = Network::curl($hcard_url);
 968                 if ($curlResult->isTimeout()) {
 969                         self::$istimeout = true;
 970                         return false;
 971                 }
 972                 $content = $curlResult->getBody();
 973                 if (!$content) {
 974                         return false;
 975                 }
 976
 977                 $doc = new DOMDocument();
 978                 if (!@$doc->loadHTML($content)) {
 979                         return false;
 980                 }
 981
 982                 $xpath = new DomXPath($doc);
 983
 984                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 985                 if (!is_object($vcards)) {
 986                         return false;
 987                 }
 988
 989                 if (!isset($data["baseurl"])) {
 990                         $data["baseurl"] = "";
 991                 }
 992
 993                 if ($vcards->length > 0) {
 994                         $vcard = $vcards->item(0);
 995
 996                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 997                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 998                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 999                         if (($search->length > 0) && empty($data["guid"])) {
1000                                 $data["guid"] = $search->item(0)->nodeValue;
1001                         }
1002
1003                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1004                         if ($search->length > 0) {
1005                                 $data["nick"] = $search->item(0)->nodeValue;
1006                         }
1007
1008                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1009                         if ($search->length > 0) {
1010                                 $data["name"] = $search->item(0)->nodeValue;
1011                         }
1012
1013                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1014                         if ($search->length > 0) {
1015                                 $data["searchable"] = $search->item(0)->nodeValue;
1016                         }
1017
1018                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1019                         if ($search->length > 0) {
1020                                 $data["pubkey"] = $search->item(0)->nodeValue;
1021                                 if (strstr($data["pubkey"], 'RSA ')) {
1022                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1023                                 }
1024                         }
1025
1026                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1027                         if ($search->length > 0) {
1028                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1029                         }
1030                 }
1031
1032                 $avatar = [];
1033                 if (!empty($vcard)) {
1034                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1035                         foreach ($photos as $photo) {
1036                                 $attr = [];
1037                                 foreach ($photo->attributes as $attribute) {
1038                                         $attr[$attribute->name] = trim($attribute->value);
1039                                 }
1040
1041                                 if (isset($attr["src"]) && isset($attr["width"])) {
1042                                         $avatar[$attr["width"]] = $attr["src"];
1043                                 }
1044
1045                                 // We don't have a width. So we just take everything that we got.
1046                                 // This is a Hubzilla workaround which doesn't send a width.
1047                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1048                                         $avatar[] = $attr["src"];
1049                                 }
1050                         }
1051                 }
1052
1053                 if (sizeof($avatar)) {
1054                         ksort($avatar);
1055                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1056                 }
1057
1058                 if ($dfrn) {
1059                         // Poll DFRN specific data
1060                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1061                         if ($search->length > 0) {
1062                                 foreach ($search as $link) {
1063                                         //$data["request"] = $search->item(0)->nodeValue;
1064                                         $attr = [];
1065                                         foreach ($link->attributes as $attribute) {
1066                                                 $attr[$attribute->name] = trim($attribute->value);
1067                                         }
1068
1069                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1070                                 }
1071                         }
1072
1073                         // Older Friendica versions had used the "uid" field differently than newer versions
1074                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1075                                 unset($data["guid"]);
1076                         }
1077                 }
1078
1079
1080                 return $data;
1081         }
1082
1083         /**
1084          * @brief Check for Diaspora contact
1085          *
1086          * @param array $webfinger Webfinger data
1087          *
1088          * @return array Diaspora data
1089          * @throws HTTPException\InternalServerErrorException
1090          */
1091         private static function diaspora($webfinger)
1092         {
1093                 $hcard_url = "";
1094                 $data = [];
1095                 // The array is reversed to take into account the order of preference for same-rel links
1096                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1097                 foreach (array_reverse($webfinger["links"]) as $link) {
1098                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1099                                 $hcard_url = $link["href"];
1100                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1101                                 $data["baseurl"] = trim($link["href"], '/');
1102                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1103                                 $data["guid"] = $link["href"];
1104                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1105                                 $data["url"] = $link["href"];
1106                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1107                                 $data["poll"] = $link["href"];
1108                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1109                                 $data["poco"] = $link["href"];
1110                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1111                                 $data["notify"] = $link["href"];
1112                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1113                                 $data["pubkey"] = base64_decode($link["href"]);
1114
1115                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1116                                 if (strstr($data["pubkey"], 'RSA ')) {
1117                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1118                                 }
1119                         }
1120                 }
1121
1122                 if (!isset($data["url"]) || ($hcard_url == "")) {
1123                         return false;
1124                 }
1125
1126                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1127                         foreach ($webfinger["aliases"] as $alias) {
1128                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1129                                         $data["alias"] = $alias;
1130                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1131                                         $data["addr"] = substr($alias, 5);
1132                                 }
1133                         }
1134                 }
1135
1136                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1137                         $data["addr"] = substr($webfinger["subject"], 5);
1138                 }
1139
1140                 // Fetch further information from the hcard
1141                 $data = self::pollHcard($hcard_url, $data);
1142
1143                 if (!$data) {
1144                         return false;
1145                 }
1146
1147                 if (isset($data["url"])
1148                         && isset($data["guid"])
1149                         && isset($data["baseurl"])
1150                         && isset($data["pubkey"])
1151                         && ($hcard_url != "")
1152                 ) {
1153                         $data["network"] = Protocol::DIASPORA;
1154
1155                         // The Diaspora handle must always be lowercase
1156                         if (!empty($data["addr"])) {
1157                                 $data["addr"] = strtolower($data["addr"]);
1158                         }
1159
1160                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1161                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1162                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1163                 } else {
1164                         return false;
1165                 }
1166
1167                 return $data;
1168         }
1169
1170         /**
1171          * @brief Check for OStatus contact
1172          *
1173          * @param array $webfinger Webfinger data
1174          * @param bool  $short     Short detection mode
1175          *
1176          * @return array|bool OStatus data or "false" on error or "true" on short mode
1177          * @throws HTTPException\InternalServerErrorException
1178          */
1179         private static function ostatus($webfinger, $short = false)
1180         {
1181                 $data = [];
1182
1183                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1184                         foreach ($webfinger["aliases"] as $alias) {
1185                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1186                                         $data["addr"] = str_replace('acct:', '', $alias);
1187                                 }
1188                         }
1189                 }
1190
1191                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1192                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1193                 ) {
1194                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1195                 }
1196
1197                 if (is_array($webfinger["links"])) {
1198                         // The array is reversed to take into account the order of preference for same-rel links
1199                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1200                         foreach (array_reverse($webfinger["links"]) as $link) {
1201                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1202                                         && (defaults($link, "type", "") == "text/html")
1203                                         && ($link["href"] != "")
1204                                 ) {
1205                                         $data["url"] = $link["href"];
1206                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1207                                         $data["notify"] = $link["href"];
1208                                 } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1209                                         $data["poll"] = $link["href"];
1210                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1211                                         $pubkey = $link["href"];
1212
1213                                         if (substr($pubkey, 0, 5) === 'data:') {
1214                                                 if (strstr($pubkey, ',')) {
1215                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1216                                                 } else {
1217                                                         $pubkey = substr($pubkey, 5);
1218                                                 }
1219                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1220                                                 $curlResult = Network::curl($pubkey);
1221                                                 if ($curlResult->isTimeout()) {
1222                                                         self::$istimeout = true;
1223                                                         return false;
1224                                                 }
1225                                                 $pubkey = $curlResult->getBody();
1226                                         }
1227
1228                                         $key = explode(".", $pubkey);
1229
1230                                         if (sizeof($key) >= 3) {
1231                                                 $m = Strings::base64UrlDecode($key[1]);
1232                                                 $e = Strings::base64UrlDecode($key[2]);
1233                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1234                                         }
1235                                 }
1236                         }
1237                 }
1238
1239                 if (isset($data["notify"]) && isset($data["pubkey"])
1240                         && isset($data["poll"])
1241                         && isset($data["url"])
1242                 ) {
1243                         $data["network"] = Protocol::OSTATUS;
1244                 } else {
1245                         return false;
1246                 }
1247
1248                 if ($short) {
1249                         return true;
1250                 }
1251
1252                 // Fetch all additional data from the feed
1253                 $curlResult = Network::curl($data["poll"]);
1254                 if ($curlResult->isTimeout()) {
1255                         self::$istimeout = true;
1256                         return false;
1257                 }
1258                 $feed = $curlResult->getBody();
1259                 $dummy1 = null;
1260                 $dummy2 = null;
1261                 $dummy2 = null;
1262                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1263                 if (!$feed_data) {
1264                         return false;
1265                 }
1266
1267                 if (!empty($feed_data["header"]["author-name"])) {
1268                         $data["name"] = $feed_data["header"]["author-name"];
1269                 }
1270                 if (!empty($feed_data["header"]["author-nick"])) {
1271                         $data["nick"] = $feed_data["header"]["author-nick"];
1272                 }
1273                 if (!empty($feed_data["header"]["author-avatar"])) {
1274                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1275                 }
1276                 if (!empty($feed_data["header"]["author-id"])) {
1277                         $data["alias"] = $feed_data["header"]["author-id"];
1278                 }
1279                 if (!empty($feed_data["header"]["author-location"])) {
1280                         $data["location"] = $feed_data["header"]["author-location"];
1281                 }
1282                 if (!empty($feed_data["header"]["author-about"])) {
1283                         $data["about"] = $feed_data["header"]["author-about"];
1284                 }
1285                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1286                 // So we take the value that we just fetched, although the other one worked as well
1287                 if (!empty($feed_data["header"]["author-link"])) {
1288                         $data["url"] = $feed_data["header"]["author-link"];
1289                 }
1290
1291                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1292                         $data['url'] = $data["alias"];
1293                         $data["alias"] = '';
1294                 }
1295
1296                 /// @todo Fetch location and "about" from the feed as well
1297                 return $data;
1298         }
1299
1300         /**
1301          * @brief Fetch data from a pump.io profile page
1302          *
1303          * @param string $profile_link Link to the profile page
1304          *
1305          * @return array profile data
1306          */
1307         private static function pumpioProfileData($profile_link)
1308         {
1309                 $doc = new DOMDocument();
1310                 if (!@$doc->loadHTMLFile($profile_link)) {
1311                         return false;
1312                 }
1313
1314                 $xpath = new DomXPath($doc);
1315
1316                 $data = [];
1317
1318                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1319
1320                 if ($data["name"] == '') {
1321                         // This is ugly - but pump.io doesn't seem to know a better way for it
1322                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1323                         $pos = strpos($data["name"], chr(10));
1324                         if ($pos) {
1325                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1326                         }
1327                 }
1328
1329                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1330
1331                 if ($data["location"] == '') {
1332                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1333                 }
1334
1335                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1336
1337                 if ($data["about"] == '') {
1338                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1339                 }
1340
1341                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1342                 if (!$avatar) {
1343                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1344                 }
1345                 if ($avatar) {
1346                         foreach ($avatar->attributes as $attribute) {
1347                                 if ($attribute->name == "src") {
1348                                         $data["photo"] = trim($attribute->value);
1349                                 }
1350                         }
1351                 }
1352
1353                 return $data;
1354         }
1355
1356         /**
1357          * @brief Check for pump.io contact
1358          *
1359          * @param array $webfinger Webfinger data
1360          *
1361          * @param       $addr
1362          * @return array pump.io data
1363          */
1364         private static function pumpio($webfinger, $addr)
1365         {
1366                 $data = [];
1367                 // The array is reversed to take into account the order of preference for same-rel links
1368                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1369                 foreach (array_reverse($webfinger["links"]) as $link) {
1370                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1371                                 && (defaults($link, "type", "") == "text/html")
1372                                 && ($link["href"] != "")
1373                         ) {
1374                                 $data["url"] = $link["href"];
1375                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1376                                 $data["notify"] = $link["href"];
1377                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1378                                 $data["poll"] = $link["href"];
1379                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1380                                 $data["dialback"] = $link["href"];
1381                         }
1382                 }
1383                 if (isset($data["poll"]) && isset($data["notify"])
1384                         && isset($data["dialback"])
1385                         && isset($data["url"])
1386                 ) {
1387                         // by now we use these fields only for the network type detection
1388                         // So we unset all data that isn't used at the moment
1389                         unset($data["dialback"]);
1390
1391                         $data["network"] = Protocol::PUMPIO;
1392                 } else {
1393                         return false;
1394                 }
1395
1396                 $profile_data = self::pumpioProfileData($data["url"]);
1397
1398                 if (!$profile_data) {
1399                         return false;
1400                 }
1401
1402                 $data = array_merge($data, $profile_data);
1403
1404                 if (($addr != '') && ($data['name'] != '')) {
1405                         $name = trim(str_replace($addr, '', $data['name']));
1406                         if ($name != '') {
1407                                 $data['name'] = $name;
1408                         }
1409                 }
1410
1411                 return $data;
1412         }
1413
1414         /**
1415          * @brief Check page for feed link
1416          *
1417          * @param string $url Page link
1418          *
1419          * @return string feed link
1420          */
1421         private static function getFeedLink($url)
1422         {
1423                 $doc = new DOMDocument();
1424
1425                 if (!@$doc->loadHTMLFile($url)) {
1426                         return false;
1427                 }
1428
1429                 $xpath = new DomXPath($doc);
1430
1431                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1432                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1433                 if (!is_object($feeds)) {
1434                         return false;
1435                 }
1436
1437                 if ($feeds->length == 0) {
1438                         return false;
1439                 }
1440
1441                 $feed_url = "";
1442
1443                 foreach ($feeds as $feed) {
1444                         $attr = [];
1445                         foreach ($feed->attributes as $attribute) {
1446                                 $attr[$attribute->name] = trim($attribute->value);
1447                         }
1448
1449                         if ($feed_url == "") {
1450                                 $feed_url = $attr["href"];
1451                         }
1452                 }
1453
1454                 return $feed_url;
1455         }
1456
1457         /**
1458          * @brief Check for feed contact
1459          *
1460          * @param string  $url   Profile link
1461          * @param boolean $probe Do a probe if the page contains a feed link
1462          *
1463          * @return array feed data
1464          * @throws HTTPException\InternalServerErrorException
1465          */
1466         private static function feed($url, $probe = true)
1467         {
1468                 $curlResult = Network::curl($url);
1469                 if ($curlResult->isTimeout()) {
1470                         self::$istimeout = true;
1471                         return false;
1472                 }
1473                 $feed = $curlResult->getBody();
1474                 $dummy1 = $dummy2 = $dummy3 = null;
1475                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1476
1477                 if (!$feed_data) {
1478                         if (!$probe) {
1479                                 return false;
1480                         }
1481
1482                         $feed_url = self::getFeedLink($url);
1483
1484                         if (!$feed_url) {
1485                                 return false;
1486                         }
1487
1488                         return self::feed($feed_url, false);
1489                 }
1490
1491                 if (!empty($feed_data["header"]["author-name"])) {
1492                         $data["name"] = $feed_data["header"]["author-name"];
1493                 }
1494
1495                 if (!empty($feed_data["header"]["author-nick"])) {
1496                         $data["nick"] = $feed_data["header"]["author-nick"];
1497                 }
1498
1499                 if (!empty($feed_data["header"]["author-avatar"])) {
1500                         $data["photo"] = $feed_data["header"]["author-avatar"];
1501                 }
1502
1503                 if (!empty($feed_data["header"]["author-id"])) {
1504                         $data["alias"] = $feed_data["header"]["author-id"];
1505                 }
1506
1507                 $data["url"] = $url;
1508                 $data["poll"] = $url;
1509
1510                 if (!empty($feed_data["header"]["author-link"])) {
1511                         $data["baseurl"] = $feed_data["header"]["author-link"];
1512                 } else {
1513                         $data["baseurl"] = $data["url"];
1514                 }
1515
1516                 $data["network"] = Protocol::FEED;
1517
1518                 return $data;
1519         }
1520
1521         /**
1522          * @brief Check for mail contact
1523          *
1524          * @param string  $uri Profile link
1525          * @param integer $uid User ID
1526          *
1527          * @return array mail data
1528          * @throws \Exception
1529          */
1530         private static function mail($uri, $uid)
1531         {
1532                 if (!Network::isEmailDomainValid($uri)) {
1533                         return false;
1534                 }
1535
1536                 if ($uid == 0) {
1537                         return false;
1538                 }
1539
1540                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1541
1542                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1543                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1544                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1545
1546                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1547                         return false;
1548                 }
1549
1550                 $mailbox = Email::constructMailboxName($mailacct);
1551                 $password = '';
1552                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1553                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1554                 if (!$mbox) {
1555                         return false;
1556                 }
1557
1558                 $msgs = Email::poll($mbox, $uri);
1559                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1560
1561                 if (!count($msgs)) {
1562                         return false;
1563                 }
1564
1565                 $phost = substr($uri, strpos($uri, '@') + 1);
1566
1567                 $data = [];
1568                 $data["addr"]    = $uri;
1569                 $data["network"] = Protocol::MAIL;
1570                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1571                 $data["nick"]    = $data["name"];
1572                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1573                 $data["url"]     = 'mailto:'.$uri;
1574                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1575                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1576
1577                 $x = Email::messageMeta($mbox, $msgs[0]);
1578                 if (stristr($x[0]->from, $uri)) {
1579                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1580                 } elseif (stristr($x[0]->to, $uri)) {
1581                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1582                 }
1583                 if (isset($adr)) {
1584                         foreach ($adr as $feadr) {
1585                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1586                                         &&(strcasecmp($feadr->host, $phost) == 0)
1587                                         && (strlen($feadr->personal))
1588                                 ) {
1589                                         $personal = imap_mime_header_decode($feadr->personal);
1590                                         $data["name"] = "";
1591                                         foreach ($personal as $perspart) {
1592                                                 if ($perspart->charset != "default") {
1593                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1594                                                 } else {
1595                                                         $data["name"] .= $perspart->text;
1596                                                 }
1597                                         }
1598
1599                                         $data["name"] = Strings::escapeTags($data["name"]);
1600                                 }
1601                         }
1602                 }
1603                 if (!empty($mbox)) {
1604                         imap_close($mbox);
1605                 }
1606                 return $data;
1607         }
1608
1609         /**
1610          * @brief Mix two paths together to possibly fix missing parts
1611          *
1612          * @param string $avatar Path to the avatar
1613          * @param string $base   Another path that is hopefully complete
1614          *
1615          * @return string fixed avatar path
1616          * @throws \Exception
1617          */
1618         public static function fixAvatar($avatar, $base)
1619         {
1620                 $base_parts = parse_url($base);
1621
1622                 // Remove all parts that could create a problem
1623                 unset($base_parts['path']);
1624                 unset($base_parts['query']);
1625                 unset($base_parts['fragment']);
1626
1627                 $avatar_parts = parse_url($avatar);
1628
1629                 // Now we mix them
1630                 $parts = array_merge($base_parts, $avatar_parts);
1631
1632                 // And put them together again
1633                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1634                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1635                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1636                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1637                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1638                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1639
1640                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1641
1642                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1643
1644                 return $fixed;
1645         }
1646 }