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\Contact;
  21 use Friendica\Model\Profile;
  22 use Friendica\Protocol\ActivityNamespace;
  23 use Friendica\Protocol\ActivityPub;
  24 use Friendica\Protocol\Email;
  25 use Friendica\Protocol\Feed;
  26 use Friendica\Util\Crypto;
  27 use Friendica\Util\Network;
  28 use Friendica\Util\Strings;
  29 use Friendica\Util\XML;
  30
  31 /**
  32  * @brief This class contain functions for probing URL
  33  *
  34  */
  35 class Probe
  36 {
  37         private static $baseurl;
  38         private static $istimeout;
  39
  40         /**
  41          * @brief Rearrange the array so that it always has the same order
  42          *
  43          * @param array $data Unordered data
  44          *
  45          * @return array Ordered data
  46          */
  47         private static function rearrangeData($data)
  48         {
  49                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
  50                                 "community", "keywords", "location", "about", "gender", "hide",
  51                                 "batch", "notify", "poll", "request", "confirm", "poco",
  52                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  53                                 "priority", "network", "pubkey", "baseurl"];
  54
  55                 $newdata = [];
  56                 foreach ($fields as $field) {
  57                         if (isset($data[$field])) {
  58                                 $newdata[$field] = $data[$field];
  59                         } else {
  60                                 $newdata[$field] = "";
  61                         }
  62                 }
  63
  64                 // We don't use the "priority" field anymore and replace it with a dummy.
  65                 $newdata["priority"] = 0;
  66
  67                 return $newdata;
  68         }
  69
  70         /**
  71          * @brief Check if the hostname belongs to the own server
  72          *
  73          * @param string $host The hostname that is to be checked
  74          *
  75          * @return bool Does the testes hostname belongs to the own server?
  76          */
  77         private static function ownHost($host)
  78         {
  79                 $own_host = \get_app()->getHostName();
  80
  81                 $parts = parse_url($host);
  82
  83                 if (!isset($parts['scheme'])) {
  84                         $parts = parse_url('http://'.$host);
  85                 }
  86
  87                 if (!isset($parts['host'])) {
  88                         return false;
  89                 }
  90                 return $parts['host'] == $own_host;
  91         }
  92
  93         /**
  94          * @brief Probes for webfinger path via "host-meta"
  95          *
  96          * We have to check if the servers in the future still will offer this.
  97          * It seems as if it was dropped from the standard.
  98          *
  99          * @param string $host The host part of an url
 100          *
 101          * @return array with template and type of the webfinger template for JSON or XML
 102          * @throws HTTPException\InternalServerErrorException
 103          */
 104         private static function hostMeta($host)
 105         {
 106                 // Reset the static variable
 107                 self::$baseurl = '';
 108
 109                 $ssl_url = "https://".$host."/.well-known/host-meta";
 110                 $url = "http://".$host."/.well-known/host-meta";
 111
 112                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 113
 114                 Logger::log("Probing for ".$host, Logger::DEBUG);
 115                 $xrd = null;
 116
 117                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 118                 if ($curlResult->isSuccess()) {
 119                         $xml = $curlResult->getBody();
 120                         $xrd = XML::parseString($xml, false);
 121                         $host_url = 'https://'.$host;
 122                 }
 123
 124                 if (!is_object($xrd)) {
 125                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 126                         if ($curlResult->isTimeout()) {
 127                                 Logger::log("Probing timeout for " . $url, Logger::DEBUG);
 128                                 self::$istimeout = true;
 129                                 return false;
 130                         }
 131                         $xml = $curlResult->getBody();
 132                         $xrd = XML::parseString($xml, false);
 133                         $host_url = 'http://'.$host;
 134                 }
 135                 if (!is_object($xrd)) {
 136                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
 137                         return [];
 138                 }
 139
 140                 $links = XML::elementToArray($xrd);
 141                 if (!isset($links["xrd"]["link"])) {
 142                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
 143                         return [];
 144                 }
 145
 146                 $lrdd = ['application/jrd+json' => $host_url . '/.well-known/webfinger?resource={uri}'];
 147
 148                 foreach ($links["xrd"]["link"] as $value => $link) {
 149                         if (!empty($link["@attributes"])) {
 150                                 $attributes = $link["@attributes"];
 151                         } elseif ($value == "@attributes") {
 152                                 $attributes = $link;
 153                         } else {
 154                                 continue;
 155                         }
 156
 157                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 158                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 159
 160                                 $lrdd[$type] = $attributes["template"];
 161                         }
 162                 }
 163
 164                 self::$baseurl = $host_url;
 165
 166                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
 167
 168                 return $lrdd;
 169         }
 170
 171         /**
 172          * @brief Perform Webfinger lookup and return DFRN data
 173          *
 174          * Given an email style address, perform webfinger lookup and
 175          * return the resulting DFRN profile URL, or if no DFRN profile URL
 176          * is located, returns an OStatus subscription template (prefixed
 177          * with the string 'stat:' to identify it as on OStatus template).
 178          * If this isn't an email style address just return $webbie.
 179          * Return an empty string if email-style addresses but webfinger fails,
 180          * or if the resultant personal XRD doesn't contain a supported
 181          * subscription/friend-request attribute.
 182          *
 183          * amended 7/9/2011 to return an hcard which could save potentially loading
 184          * a lengthy content page to scrape dfrn attributes
 185          *
 186          * @param string $webbie    Address that should be probed
 187          * @param string $hcard_url Link to the hcard - is returned by reference
 188          *
 189          * @return string profile link
 190          * @throws HTTPException\InternalServerErrorException
 191          */
 192         public static function webfingerDfrn($webbie, &$hcard_url)
 193         {
 194                 $profile_link = '';
 195
 196                 $links = self::lrdd($webbie);
 197                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
 198                 if (!empty($links) && is_array($links)) {
 199                         foreach ($links as $link) {
 200                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
 201                                         $profile_link = $link['@attributes']['href'];
 202                                 }
 203                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
 204                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 205                                 }
 206                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 207                                         $hcard_url = $link['@attributes']['href'];
 208                                 }
 209                         }
 210                 }
 211                 return $profile_link;
 212         }
 213
 214         /**
 215          * @brief Check an URI for LRDD data
 216          *
 217          * this is a replacement for the "lrdd" function.
 218          * It isn't used in this class and has some redundancies in the code.
 219          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 220          *
 221          * @param string $uri Address that should be probed
 222          *
 223          * @return array uri data
 224          * @throws HTTPException\InternalServerErrorException
 225          */
 226         public static function lrdd($uri)
 227         {
 228                 $lrdd = self::hostMeta($uri);
 229                 $webfinger = null;
 230
 231                 if (is_bool($lrdd)) {
 232                         return [];
 233                 }
 234
 235                 if (!$lrdd) {
 236                         $parts = @parse_url($uri);
 237                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 238                                 return [];
 239                         }
 240
 241                         $host = $parts["host"];
 242                         if (!empty($parts["port"])) {
 243                                 $host .= ':'.$parts["port"];
 244                         }
 245
 246                         $path_parts = explode("/", trim($parts["path"], "/"));
 247
 248                         $nick = array_pop($path_parts);
 249
 250                         do {
 251                                 $lrdd = self::hostMeta($host);
 252                                 $host .= "/".array_shift($path_parts);
 253                         } while (!$lrdd && (sizeof($path_parts) > 0));
 254                 }
 255
 256                 if (!$lrdd) {
 257                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
 258                         return [];
 259                 }
 260
 261                 foreach ($lrdd as $type => $template) {
 262                         if ($webfinger) {
 263                                 continue;
 264                         }
 265
 266                         $path = str_replace('{uri}', urlencode($uri), $template);
 267                         $webfinger = self::webfinger($path, $type);
 268
 269                         if (!$webfinger && (strstr($uri, "@"))) {
 270                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 271                                 $webfinger = self::webfinger($path, $type);
 272                         }
 273
 274                         // Special treatment for Mastodon
 275                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 276                         // But the webfinger for this format fails.
 277                         if (!$webfinger && !empty($nick)) {
 278                                 // Mastodon uses a "@" as prefix for usernames in their url format
 279                                 $nick = ltrim($nick, '@');
 280
 281                                 $addr = $nick."@".$host;
 282
 283                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 284                                 $webfinger = self::webfinger($path, $type);
 285                         }
 286                 }
 287
 288                 if (!is_array($webfinger["links"])) {
 289                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
 290                         return false;
 291                 }
 292
 293                 $data = [];
 294
 295                 foreach ($webfinger["links"] as $link) {
 296                         $data[] = ["@attributes" => $link];
 297                 }
 298
 299                 if (is_array($webfinger["aliases"])) {
 300                         foreach ($webfinger["aliases"] as $alias) {
 301                                 $data[] = ["@attributes" =>
 302                                                         ["rel" => "alias",
 303                                                                 "href" => $alias]];
 304                         }
 305                 }
 306
 307                 return $data;
 308         }
 309
 310         /**
 311          * @brief Fetch information (protocol endpoints and user information) about a given uri
 312          *
 313          * @param string  $uri     Address that should be probed
 314          * @param string  $network Test for this specific network
 315          * @param integer $uid     User ID for the probe (only used for mails)
 316          * @param boolean $cache   Use cached values?
 317          *
 318          * @return array uri data
 319          * @throws HTTPException\InternalServerErrorException
 320          * @throws \ImagickException
 321          */
 322         public static function uri($uri, $network = '', $uid = -1, $cache = true)
 323         {
 324                 if ($cache) {
 325                         $result = Cache::get('Probe::uri:' . $network . ':' . $uri);
 326                         if (!is_null($result)) {
 327                                 return $result;
 328                         }
 329                 }
 330
 331                 if ($uid == -1) {
 332                         $uid = local_user();
 333                 }
 334
 335                 self::$istimeout = false;
 336
 337                 if ($network != Protocol::ACTIVITYPUB) {
 338                         $data = self::detect($uri, $network, $uid);
 339                 } else {
 340                         $data = null;
 341                 }
 342
 343                 // When the previous detection process had got a time out
 344                 // we could falsely detect a Friendica profile as AP profile.
 345                 if (!self::$istimeout) {
 346                         $ap_profile = ActivityPub::probeProfile($uri);
 347
 348                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 349                                 $data = $ap_profile;
 350                         } elseif (!empty($ap_profile)) {
 351                                 $ap_profile['batch'] = '';
 352                                 $data = array_merge($ap_profile, $data);
 353                         }
 354                 } else {
 355                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 356                 }
 357
 358                 if (!isset($data['url'])) {
 359                         $data['url'] = $uri;
 360                 }
 361
 362                 if (!empty($data['photo']) && !empty($data['baseurl'])) {
 363                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl']), Strings::normaliseLink($data['photo']));
 364                 } elseif (empty($data['photo'])) {
 365                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 366                 }
 367
 368                 if (empty($data['name'])) {
 369                         if (!empty($data['nick'])) {
 370                                 $data['name'] = $data['nick'];
 371                         }
 372
 373                         if (empty($data['name'])) {
 374                                 $data['name'] = $data['url'];
 375                         }
 376                 }
 377
 378                 if (empty($data['nick'])) {
 379                         $data['nick'] = strtolower($data['name']);
 380
 381                         if (strpos($data['nick'], ' ')) {
 382                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 383                         }
 384                 }
 385
 386                 if (!empty(self::$baseurl)) {
 387                         $data['baseurl'] = self::$baseurl;
 388                 }
 389
 390                 if (empty($data['network'])) {
 391                         $data['network'] = Protocol::PHANTOM;
 392                 }
 393
 394                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 395                         $data['hide'] = self::getHideStatus($data['url']);
 396                 }
 397
 398                 $data = self::rearrangeData($data);
 399
 400                 // Only store into the cache if the value seems to be valid
 401                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 402                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 403                 }
 404
 405                 return $data;
 406         }
 407
 408
 409         /**
 410          * Fetches the "hide" status from the profile
 411          *
 412          * @param string $url URL of the profile
 413          *
 414          * @return boolean "hide" status
 415          */
 416         private static function getHideStatus($url)
 417         {
 418                 $curlResult = Network::curl($url);
 419                 if (!$curlResult->isSuccess()) {
 420                         return false;
 421                 }
 422
 423                 // If the file is too large then exit
 424                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 425                         return false;
 426                 }
 427
 428                 // If it isn't a HTML file then exit
 429                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 430                         return false;
 431                 }
 432
 433                 $body = $curlResult->getBody();
 434
 435                 $doc = new DOMDocument();
 436                 @$doc->loadHTML($body);
 437
 438                 $xpath = new DOMXPath($doc);
 439
 440                 $list = $xpath->query('//meta[@name]');
 441                 foreach ($list as $node) {
 442                         $meta_tag = [];
 443                         if ($node->attributes->length) {
 444                                 foreach ($node->attributes as $attribute) {
 445                                         $meta_tag[$attribute->name] = $attribute->value;
 446                                 }
 447                         }
 448
 449                         if (empty($meta_tag['content'])) {
 450                                 continue;
 451                         }
 452
 453                         $content = strtolower(trim($meta_tag['content']));
 454
 455                         switch (strtolower(trim($meta_tag['name']))) {
 456                                 case 'dfrn-global-visibility':
 457                                         if ($content == 'false') {
 458                                                 return true;
 459                                         }
 460                                         break;
 461                                 case 'robots':
 462                                         if (strpos($content, 'noindex') !== false) {
 463                                                 return true;
 464                                         }
 465                                         break;
 466                         }
 467                 }
 468
 469                 return false;
 470         }
 471
 472         /**
 473          * @brief Checks if a profile url should be OStatus but only provides partial information
 474          *
 475          * @param array  $webfinger Webfinger data
 476          * @param string $lrdd      Path template for webfinger request
 477          * @param string $type      type
 478          *
 479          * @return array fixed webfinger data
 480          * @throws HTTPException\InternalServerErrorException
 481          */
 482         private static function fixOStatus($webfinger, $lrdd, $type)
 483         {
 484                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 485                         return $webfinger;
 486                 }
 487
 488                 $is_ostatus = false;
 489                 $has_key = false;
 490
 491                 foreach ($webfinger['links'] as $link) {
 492                         if ($link['rel'] == ActivityNamespace::OSTATUSSUB) {
 493                                 $is_ostatus = true;
 494                         }
 495                         if ($link['rel'] == 'magic-public-key') {
 496                                 $has_key = true;
 497                         }
 498                 }
 499
 500                 if (!$is_ostatus || $has_key) {
 501                         return $webfinger;
 502                 }
 503
 504                 $url = Network::switchScheme($webfinger['subject']);
 505                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 506                 $webfinger2 = self::webfinger($path, $type);
 507
 508                 // Is the new webfinger detectable as OStatus?
 509                 if (self::ostatus($webfinger2, true)) {
 510                         $webfinger = $webfinger2;
 511                 }
 512
 513                 return $webfinger;
 514         }
 515
 516         /**
 517          * @brief Fetch information (protocol endpoints and user information) about a given uri
 518          *
 519          * This function is only called by the "uri" function that adds caching and rearranging of data.
 520          *
 521          * @param string  $uri     Address that should be probed
 522          * @param string  $network Test for this specific network
 523          * @param integer $uid     User ID for the probe (only used for mails)
 524          *
 525          * @return array uri data
 526          * @throws HTTPException\InternalServerErrorException
 527          */
 528         private static function detect($uri, $network, $uid)
 529         {
 530                 $parts = parse_url($uri);
 531
 532                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 533                         $host = $parts["host"];
 534                         if (!empty($parts["port"])) {
 535                                 $host .= ':'.$parts["port"];
 536                         }
 537
 538                         if ($host == 'twitter.com') {
 539                                 return self::twitter($uri);
 540                         }
 541                         $lrdd = self::hostMeta($host);
 542
 543                         if (is_bool($lrdd)) {
 544                                 return [];
 545                         }
 546
 547                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
 548
 549                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 550                                 $host .= "/".array_shift($path_parts);
 551                                 $lrdd = self::hostMeta($host);
 552                         }
 553                         if (!$lrdd) {
 554                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 555                                 return self::feed($uri);
 556                         }
 557                         $nick = array_pop($path_parts);
 558
 559                         // Mastodon uses a "@" as prefix for usernames in their url format
 560                         $nick = ltrim($nick, '@');
 561
 562                         $addr = $nick."@".$host;
 563                 } elseif (strstr($uri, '@')) {
 564                         // If the URI starts with "mailto:" then jump directly to the mail detection
 565                         if (strpos($uri, 'mailto:') !== false) {
 566                                 $uri = str_replace('mailto:', '', $uri);
 567                                 return self::mail($uri, $uid);
 568                         }
 569
 570                         if ($network == Protocol::MAIL) {
 571                                 return self::mail($uri, $uid);
 572                         }
 573                         // Remove "acct:" from the URI
 574                         $uri = str_replace('acct:', '', $uri);
 575
 576                         $host = substr($uri, strpos($uri, '@') + 1);
 577                         $nick = substr($uri, 0, strpos($uri, '@'));
 578
 579                         if (strpos($uri, '@twitter.com')) {
 580                                 return self::twitter($uri);
 581                         }
 582                         $lrdd = self::hostMeta($host);
 583
 584                         if (is_bool($lrdd)) {
 585                                 return [];
 586                         }
 587
 588                         if (!$lrdd) {
 589                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 590                                 return self::mail($uri, $uid);
 591                         }
 592                         $addr = $uri;
 593                 } else {
 594                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 595                         return false;
 596                 }
 597
 598                 $webfinger = false;
 599
 600                 /// @todo Do we need the prefix "acct:" or "acct://"?
 601
 602                 foreach ($lrdd as $type => $template) {
 603                         if ($webfinger) {
 604                                 continue;
 605                         }
 606
 607                         // At first try it with the given uri
 608                         $path = str_replace('{uri}', urlencode($uri), $template);
 609                         $webfinger = self::webfinger($path, $type);
 610
 611                         // Fix possible problems with GNU Social probing to wrong scheme
 612                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 613
 614                         // We cannot be sure that the detected address was correct, so we don't use the values
 615                         if ($webfinger && ($uri != $addr)) {
 616                                 $nick = "";
 617                                 $addr = "";
 618                         }
 619
 620                         // Try webfinger with the address (user@domain.tld)
 621                         if (!$webfinger) {
 622                                 $path = str_replace('{uri}', urlencode($addr), $template);
 623                                 $webfinger = self::webfinger($path, $type);
 624                         }
 625
 626                         // Mastodon needs to have it with "acct:"
 627                         if (!$webfinger) {
 628                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 629                                 $webfinger = self::webfinger($path, $type);
 630                         }
 631                 }
 632
 633                 if (!$webfinger) {
 634                         return self::feed($uri);
 635                 }
 636
 637                 $result = false;
 638
 639                 Logger::log("Probing ".$uri, Logger::DEBUG);
 640
 641                 if (in_array($network, ["", Protocol::DFRN])) {
 642                         $result = self::dfrn($webfinger);
 643                 }
 644                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 645                         $result = self::diaspora($webfinger);
 646                 }
 647                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 648                         $result = self::ostatus($webfinger, false);
 649                 }
 650 //              if (in_array($network, ['', Protocol::ZOT])) {
 651 //                      $result = self::zot($webfinger, $result);
 652 //              }
 653                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 654                         $result = self::pumpio($webfinger, $addr);
 655                 }
 656                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 657                         $result = self::feed($uri, true);
 658                 } else {
 659                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 660                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 661                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 662                                 $result["nick"] = $nick;
 663                         }
 664
 665                         if (empty($result["addr"]) && ($addr != "")) {
 666                                 $result["addr"] = $addr;
 667                         }
 668                 }
 669
 670                 if (empty($result["network"])) {
 671                         $result["network"] = Protocol::PHANTOM;
 672                 }
 673
 674                 if (empty($result["url"])) {
 675                         $result["url"] = $uri;
 676                 }
 677
 678                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 679
 680                 if (empty($result["baseurl"]) && ($result["network"] != Protocol::PHANTOM)) {
 681                         $pos = strpos($result["url"], $host);
 682                         if ($pos) {
 683                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 684                         }
 685                 }
 686                 return $result;
 687         }
 688
 689         /**
 690          * Check for Zot contact
 691          *
 692          * @param array $webfinger Webfinger data
 693          * @param array $data      previously probed data
 694          *
 695          * @return array Zot data
 696          * @throws HTTPException\InternalServerErrorException
 697          */
 698         private static function zot($webfinger, $data)
 699         {
 700                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 701                         foreach ($webfinger["aliases"] as $alias) {
 702                                 if (substr($alias, 0, 5) == 'acct:') {
 703                                         $data["addr"] = substr($alias, 5);
 704                                 }
 705                         }
 706                 }
 707
 708                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 709                         $data["addr"] = substr($webfinger["subject"], 5);
 710                 }
 711
 712                 $zot_url = '';
 713                 foreach ($webfinger['links'] as $link) {
 714                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 715                                 $zot_url = $link['href'];
 716                         }
 717                 }
 718
 719                 if (empty($zot_url) && !empty($data['addr']) && !empty(self::$baseurl)) {
 720                         $condition = ['nurl' => Strings::normaliseLink(self::$baseurl), 'platform' => ['hubzilla']];
 721                         if (!DBA::exists('gserver', $condition)) {
 722                                 exit;
 723                         }
 724                         $zot_url = self::$baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 725                 }
 726
 727                 if (!empty($zot_url)) {
 728                         $data = self::pollZot($zot_url, $data);
 729                 }
 730
 731                 return $data;
 732         }
 733
 734         public static function pollZot($url, $data)
 735         {
 736                 $curlResult = Network::curl($url);
 737                 if ($curlResult->isTimeout()) {
 738                         return $data;
 739                 }
 740                 $content = $curlResult->getBody();
 741                 if (!$content) {
 742                         return $data;
 743                 }
 744
 745                 $json = json_decode($content, true);
 746                 if (!is_array($json)) {
 747                         return $data;
 748                 }
 749 //print_r($json);
 750                 if (empty($data['network'])) {
 751                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 752                                 $data['network'] = Protocol::ZOT;
 753                         } elseif (!isset($json['protocols'])) {
 754                                 $data['network'] = Protocol::ZOT;
 755                         }
 756                 }
 757
 758                 if (!empty($json['guid']) && empty($data['guid'])) {
 759                         $data['guid'] = $json['guid'];
 760                 }
 761                 if (!empty($json['key']) && empty($data['pubkey'])) {
 762                         $data['pubkey'] = $json['key'];
 763                 }
 764                 if (!empty($json['name'])) {
 765                         $data['name'] = $json['name'];
 766                 }
 767                 if (!empty($json['photo'])) {
 768                         $data['photo'] = $json['photo'];
 769                         if (!empty($json['photo_updated'])) {
 770                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 771                         }
 772                 }
 773                 if (!empty($json['address'])) {
 774                         $data['addr'] = $json['address'];
 775                 }
 776                 if (!empty($json['url'])) {
 777                         $data['url'] = $json['url'];
 778                 }
 779                 if (!empty($json['connections_url'])) {
 780                         $data['poco'] = $json['connections_url'];
 781                 }
 782                 if (isset($json['searchable'])) {
 783                         $data['hide'] = !$json['searchable'];
 784                 }
 785                 if (!empty($json['public_forum'])) {
 786                         $data['community'] = $json['public_forum'];
 787                         $data['account-type'] = Contact::PAGE_COMMUNITY;
 788                 }
 789
 790                 if (!empty($json['profile'])) {
 791                         $profile = $json['profile'];
 792                         if (!empty($profile['description'])) {
 793                                 $data['about'] = $profile['description'];
 794                         }
 795                         if (!empty($profile['gender'])) {
 796                                 $data['gender'] = $profile['gender'];
 797                         }
 798                         if (!empty($profile['keywords'])) {
 799                                 $keywords = implode(', ', $profile['keywords']);
 800                                 if (!empty($keywords)) {
 801                                         $data['keywords'] = $keywords;
 802                                 }
 803                         }
 804
 805                         $loc = [];
 806                         if (!empty($profile['region'])) {
 807                                 $loc['region'] = $profile['region'];
 808                         }
 809                         if (!empty($profile['country'])) {
 810                                 $loc['country-name'] = $profile['country'];
 811                         }
 812                         if (!empty($profile['hometown'])) {
 813                                 $loc['locality'] = $profile['hometown'];
 814                         }
 815                         $location = Profile::formatLocation($loc);
 816                         if (!empty($location)) {
 817                                 $data['location'] = $location;
 818                         }
 819                 }
 820
 821                 return $data;
 822         }
 823
 824         /**
 825          * @brief Perform a webfinger request.
 826          *
 827          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 828          *
 829          * @param string $url  Address that should be probed
 830          * @param string $type type
 831          *
 832          * @return array webfinger data
 833          * @throws HTTPException\InternalServerErrorException
 834          */
 835         private static function webfinger($url, $type)
 836         {
 837                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 838
 839                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 840                 if ($curlResult->isTimeout()) {
 841                         self::$istimeout = true;
 842                         return false;
 843                 }
 844                 $data = $curlResult->getBody();
 845
 846                 $webfinger = json_decode($data, true);
 847                 if (is_array($webfinger)) {
 848                         if (!isset($webfinger["links"])) {
 849                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 850                                 return false;
 851                         }
 852                         return $webfinger;
 853                 }
 854
 855                 // If it is not JSON, maybe it is XML
 856                 $xrd = XML::parseString($data, false);
 857                 if (!is_object($xrd)) {
 858                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 859                         return false;
 860                 }
 861
 862                 $xrd_arr = XML::elementToArray($xrd);
 863                 if (!isset($xrd_arr["xrd"]["link"])) {
 864                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 865                         return false;
 866                 }
 867
 868                 $webfinger = [];
 869
 870                 if (!empty($xrd_arr["xrd"]["subject"])) {
 871                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 872                 }
 873
 874                 if (!empty($xrd_arr["xrd"]["alias"])) {
 875                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 876                 }
 877
 878                 $webfinger["links"] = [];
 879
 880                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 881                         if (!empty($data["@attributes"])) {
 882                                 $attributes = $data["@attributes"];
 883                         } elseif ($value == "@attributes") {
 884                                 $attributes = $data;
 885                         } else {
 886                                 continue;
 887                         }
 888
 889                         $webfinger["links"][] = $attributes;
 890                 }
 891                 return $webfinger;
 892         }
 893
 894         /**
 895          * @brief Poll the Friendica specific noscrape page.
 896          *
 897          * "noscrape" is a faster alternative to fetch the data from the hcard.
 898          * This functionality was originally created for the directory.
 899          *
 900          * @param string $noscrape_url Link to the noscrape page
 901          * @param array  $data         The already fetched data
 902          *
 903          * @return array noscrape data
 904          * @throws HTTPException\InternalServerErrorException
 905          */
 906         private static function pollNoscrape($noscrape_url, $data)
 907         {
 908                 $curlResult = Network::curl($noscrape_url);
 909                 if ($curlResult->isTimeout()) {
 910                         self::$istimeout = true;
 911                         return false;
 912                 }
 913                 $content = $curlResult->getBody();
 914                 if (!$content) {
 915                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 916                         return false;
 917                 }
 918
 919                 $json = json_decode($content, true);
 920                 if (!is_array($json)) {
 921                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 922                         return false;
 923                 }
 924
 925                 if (!empty($json["fn"])) {
 926                         $data["name"] = $json["fn"];
 927                 }
 928
 929                 if (!empty($json["addr"])) {
 930                         $data["addr"] = $json["addr"];
 931                 }
 932
 933                 if (!empty($json["nick"])) {
 934                         $data["nick"] = $json["nick"];
 935                 }
 936
 937                 if (!empty($json["guid"])) {
 938                         $data["guid"] = $json["guid"];
 939                 }
 940
 941                 if (!empty($json["comm"])) {
 942                         $data["community"] = $json["comm"];
 943                 }
 944
 945                 if (!empty($json["tags"])) {
 946                         $keywords = implode(", ", $json["tags"]);
 947                         if ($keywords != "") {
 948                                 $data["keywords"] = $keywords;
 949                         }
 950                 }
 951
 952                 $location = Profile::formatLocation($json);
 953                 if ($location) {
 954                         $data["location"] = $location;
 955                 }
 956
 957                 if (!empty($json["about"])) {
 958                         $data["about"] = $json["about"];
 959                 }
 960
 961                 if (!empty($json["gender"])) {
 962                         $data["gender"] = $json["gender"];
 963                 }
 964
 965                 if (!empty($json["key"])) {
 966                         $data["pubkey"] = $json["key"];
 967                 }
 968
 969                 if (!empty($json["photo"])) {
 970                         $data["photo"] = $json["photo"];
 971                 }
 972
 973                 if (!empty($json["dfrn-request"])) {
 974                         $data["request"] = $json["dfrn-request"];
 975                 }
 976
 977                 if (!empty($json["dfrn-confirm"])) {
 978                         $data["confirm"] = $json["dfrn-confirm"];
 979                 }
 980
 981                 if (!empty($json["dfrn-notify"])) {
 982                         $data["notify"] = $json["dfrn-notify"];
 983                 }
 984
 985                 if (!empty($json["dfrn-poll"])) {
 986                         $data["poll"] = $json["dfrn-poll"];
 987                 }
 988
 989                 if (isset($json["hide"])) {
 990                         $data["hide"] = (bool)$json["hide"];
 991                 } else {
 992                         $data["hide"] = false;
 993                 }
 994
 995                 return $data;
 996         }
 997
 998         /**
 999          * @brief Check for valid DFRN data
1000          *
1001          * @param array $data DFRN data
1002          *
1003          * @return int Number of errors
1004          */
1005         public static function validDfrn($data)
1006         {
1007                 $errors = 0;
1008                 if (!isset($data['key'])) {
1009                         $errors ++;
1010                 }
1011                 if (!isset($data['dfrn-request'])) {
1012                         $errors ++;
1013                 }
1014                 if (!isset($data['dfrn-confirm'])) {
1015                         $errors ++;
1016                 }
1017                 if (!isset($data['dfrn-notify'])) {
1018                         $errors ++;
1019                 }
1020                 if (!isset($data['dfrn-poll'])) {
1021                         $errors ++;
1022                 }
1023                 return $errors;
1024         }
1025
1026         /**
1027          * @brief Fetch data from a DFRN profile page and via "noscrape"
1028          *
1029          * @param string $profile_link Link to the profile page
1030          *
1031          * @return array profile data
1032          * @throws HTTPException\InternalServerErrorException
1033          * @throws \ImagickException
1034          */
1035         public static function profile($profile_link)
1036         {
1037                 $data = [];
1038
1039                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
1040
1041                 // Fetch data via noscrape - this is faster
1042                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1043                 $data = self::pollNoscrape($noscrape_url, $data);
1044
1045                 if (!isset($data["notify"])
1046                         || !isset($data["confirm"])
1047                         || !isset($data["request"])
1048                         || !isset($data["poll"])
1049                         || !isset($data["name"])
1050                         || !isset($data["photo"])
1051                 ) {
1052                         $data = self::pollHcard($profile_link, $data, true);
1053                 }
1054
1055                 $prof_data = [];
1056
1057                 if (empty($data["addr"]) || empty($data["nick"])) {
1058                         $probe_data = self::uri($profile_link);
1059                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1060                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1061                 }
1062
1063                 $prof_data["addr"]         = $data["addr"];
1064                 $prof_data["nick"]         = $data["nick"];
1065                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1066                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1067                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1068                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1069                 $prof_data["photo"]        = $data['photo']   ?? null;
1070                 $prof_data["fn"]           = $data['name']    ?? null;
1071                 $prof_data["key"]          = $data['pubkey']  ?? null;
1072
1073                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
1074
1075                 return $prof_data;
1076         }
1077
1078         /**
1079          * @brief Check for DFRN contact
1080          *
1081          * @param array $webfinger Webfinger data
1082          *
1083          * @return array DFRN data
1084          * @throws HTTPException\InternalServerErrorException
1085          */
1086         private static function dfrn($webfinger)
1087         {
1088                 $hcard_url = "";
1089                 $data = [];
1090                 // The array is reversed to take into account the order of preference for same-rel links
1091                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1092                 foreach (array_reverse($webfinger["links"]) as $link) {
1093                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1094                                 $data["network"] = Protocol::DFRN;
1095                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1096                                 $data["poll"] = $link["href"];
1097                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1098                                 $data["url"] = $link["href"];
1099                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1100                                 $hcard_url = $link["href"];
1101                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1102                                 $data["poco"] = $link["href"];
1103                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1104                                 $data["photo"] = $link["href"];
1105                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1106                                 $data["baseurl"] = trim($link["href"], '/');
1107                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1108                                 $data["guid"] = $link["href"];
1109                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1110                                 $data["pubkey"] = base64_decode($link["href"]);
1111
1112                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1113                                 if (strstr($data["pubkey"], 'RSA ')) {
1114                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1115                                 }
1116                         }
1117                 }
1118
1119                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1120                         foreach ($webfinger["aliases"] as $alias) {
1121                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1122                                         $data["url"] = $alias;
1123                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1124                                         $data["alias"] = $alias;
1125                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1126                                         $data["addr"] = substr($alias, 5);
1127                                 }
1128                         }
1129                 }
1130
1131                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1132                         $data["addr"] = substr($webfinger["subject"], 5);
1133                 }
1134
1135                 if (!isset($data["network"]) || ($hcard_url == "")) {
1136                         return false;
1137                 }
1138
1139                 // Fetch data via noscrape - this is faster
1140                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1141                 $data = self::pollNoscrape($noscrape_url, $data);
1142
1143                 if (isset($data["notify"])
1144                         && isset($data["confirm"])
1145                         && isset($data["request"])
1146                         && isset($data["poll"])
1147                         && isset($data["name"])
1148                         && isset($data["photo"])
1149                 ) {
1150                         return $data;
1151                 }
1152
1153                 $data = self::pollHcard($hcard_url, $data, true);
1154
1155                 return $data;
1156         }
1157
1158         /**
1159          * @brief Poll the hcard page (Diaspora and Friendica specific)
1160          *
1161          * @param string  $hcard_url Link to the hcard page
1162          * @param array   $data      The already fetched data
1163          * @param boolean $dfrn      Poll DFRN specific data
1164          *
1165          * @return array hcard data
1166          * @throws HTTPException\InternalServerErrorException
1167          */
1168         private static function pollHcard($hcard_url, $data, $dfrn = false)
1169         {
1170                 $curlResult = Network::curl($hcard_url);
1171                 if ($curlResult->isTimeout()) {
1172                         self::$istimeout = true;
1173                         return false;
1174                 }
1175                 $content = $curlResult->getBody();
1176                 if (!$content) {
1177                         return false;
1178                 }
1179
1180                 $doc = new DOMDocument();
1181                 if (!@$doc->loadHTML($content)) {
1182                         return false;
1183                 }
1184
1185                 $xpath = new DomXPath($doc);
1186
1187                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1188                 if (!is_object($vcards)) {
1189                         return false;
1190                 }
1191
1192                 if (!isset($data["baseurl"])) {
1193                         $data["baseurl"] = "";
1194                 }
1195
1196                 if ($vcards->length > 0) {
1197                         $vcard = $vcards->item(0);
1198
1199                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1200                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1201                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1202                         if (($search->length > 0) && empty($data["guid"])) {
1203                                 $data["guid"] = $search->item(0)->nodeValue;
1204                         }
1205
1206                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1207                         if ($search->length > 0) {
1208                                 $data["nick"] = $search->item(0)->nodeValue;
1209                         }
1210
1211                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1212                         if ($search->length > 0) {
1213                                 $data["name"] = $search->item(0)->nodeValue;
1214                         }
1215
1216                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1217                         if ($search->length > 0) {
1218                                 $data["searchable"] = $search->item(0)->nodeValue;
1219                         }
1220
1221                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1222                         if ($search->length > 0) {
1223                                 $data["pubkey"] = $search->item(0)->nodeValue;
1224                                 if (strstr($data["pubkey"], 'RSA ')) {
1225                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1226                                 }
1227                         }
1228
1229                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1230                         if ($search->length > 0) {
1231                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1232                         }
1233                 }
1234
1235                 $avatar = [];
1236                 if (!empty($vcard)) {
1237                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1238                         foreach ($photos as $photo) {
1239                                 $attr = [];
1240                                 foreach ($photo->attributes as $attribute) {
1241                                         $attr[$attribute->name] = trim($attribute->value);
1242                                 }
1243
1244                                 if (isset($attr["src"]) && isset($attr["width"])) {
1245                                         $avatar[$attr["width"]] = $attr["src"];
1246                                 }
1247
1248                                 // We don't have a width. So we just take everything that we got.
1249                                 // This is a Hubzilla workaround which doesn't send a width.
1250                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1251                                         $avatar[] = $attr["src"];
1252                                 }
1253                         }
1254                 }
1255
1256                 if (sizeof($avatar)) {
1257                         ksort($avatar);
1258                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1259                 }
1260
1261                 if ($dfrn) {
1262                         // Poll DFRN specific data
1263                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1264                         if ($search->length > 0) {
1265                                 foreach ($search as $link) {
1266                                         //$data["request"] = $search->item(0)->nodeValue;
1267                                         $attr = [];
1268                                         foreach ($link->attributes as $attribute) {
1269                                                 $attr[$attribute->name] = trim($attribute->value);
1270                                         }
1271
1272                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1273                                 }
1274                         }
1275
1276                         // Older Friendica versions had used the "uid" field differently than newer versions
1277                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1278                                 unset($data["guid"]);
1279                         }
1280                 }
1281
1282
1283                 return $data;
1284         }
1285
1286         /**
1287          * @brief Check for Diaspora contact
1288          *
1289          * @param array $webfinger Webfinger data
1290          *
1291          * @return array Diaspora data
1292          * @throws HTTPException\InternalServerErrorException
1293          */
1294         private static function diaspora($webfinger)
1295         {
1296                 $hcard_url = "";
1297
1298                 $data = [];
1299
1300                 // The array is reversed to take into account the order of preference for same-rel links
1301                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1302                 foreach (array_reverse($webfinger["links"]) as $link) {
1303                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1304                                 $hcard_url = $link["href"];
1305                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1306                                 $data["baseurl"] = trim($link["href"], '/');
1307                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1308                                 $data["guid"] = $link["href"];
1309                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1310                                 $data["url"] = $link["href"];
1311                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1312                                 $data["poll"] = $link["href"];
1313                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1314                                 $data["poco"] = $link["href"];
1315                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1316                                 $data["notify"] = $link["href"];
1317                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1318                                 $data["pubkey"] = base64_decode($link["href"]);
1319
1320                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1321                                 if (strstr($data["pubkey"], 'RSA ')) {
1322                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1323                                 }
1324                         }
1325                 }
1326
1327                 if (empty($data["url"]) || empty($hcard_url)) {
1328                         return $data;
1329                 }
1330
1331                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1332                         foreach ($webfinger["aliases"] as $alias) {
1333                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1334                                         $data["alias"] = $alias;
1335                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1336                                         $data["addr"] = substr($alias, 5);
1337                                 }
1338                         }
1339                 }
1340
1341                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1342                         $data["addr"] = substr($webfinger["subject"], 5);
1343                 }
1344
1345                 // Fetch further information from the hcard
1346                 $data = self::pollHcard($hcard_url, $data);
1347
1348                 if (!empty($data["url"])
1349                         && !empty($data["guid"])
1350                         && !empty($data["baseurl"])
1351                         && !empty($data["pubkey"])
1352                         && !empty($hcard_url)
1353                 ) {
1354                         $data["network"] = Protocol::DIASPORA;
1355
1356                         // The Diaspora handle must always be lowercase
1357                         if (!empty($data["addr"])) {
1358                                 $data["addr"] = strtolower($data["addr"]);
1359                         }
1360
1361                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1362                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1363                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1364                 }
1365
1366                 return $data;
1367         }
1368
1369         /**
1370          * @brief Check for OStatus contact
1371          *
1372          * @param array $webfinger Webfinger data
1373          * @param bool  $short     Short detection mode
1374          *
1375          * @return array|bool OStatus data or "false" on error or "true" on short mode
1376          * @throws HTTPException\InternalServerErrorException
1377          */
1378         private static function ostatus($webfinger, $short = false)
1379         {
1380                 $data = [];
1381
1382                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1383                         foreach ($webfinger["aliases"] as $alias) {
1384                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1385                                         $data["addr"] = str_replace('acct:', '', $alias);
1386                                 }
1387                         }
1388                 }
1389
1390                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1391                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1392                 ) {
1393                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1394                 }
1395
1396                 if (is_array($webfinger["links"])) {
1397                         // The array is reversed to take into account the order of preference for same-rel links
1398                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1399                         foreach (array_reverse($webfinger["links"]) as $link) {
1400                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1401                                         && (($link["type"] ?? "") == "text/html")
1402                                         && ($link["href"] != "")
1403                                 ) {
1404                                         $data["url"] = $link["href"];
1405                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1406                                         $data["notify"] = $link["href"];
1407                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1408                                         $data["poll"] = $link["href"];
1409                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1410                                         $pubkey = $link["href"];
1411
1412                                         if (substr($pubkey, 0, 5) === 'data:') {
1413                                                 if (strstr($pubkey, ',')) {
1414                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1415                                                 } else {
1416                                                         $pubkey = substr($pubkey, 5);
1417                                                 }
1418                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1419                                                 $curlResult = Network::curl($pubkey);
1420                                                 if ($curlResult->isTimeout()) {
1421                                                         self::$istimeout = true;
1422                                                         if ($short) {
1423                                                                 return false;
1424                                                         } else {
1425                                                                 return $data;
1426                                                         }
1427                                                 }
1428                                                 $pubkey = $curlResult->getBody();
1429                                         }
1430
1431                                         $key = explode(".", $pubkey);
1432
1433                                         if (sizeof($key) >= 3) {
1434                                                 $m = Strings::base64UrlDecode($key[1]);
1435                                                 $e = Strings::base64UrlDecode($key[2]);
1436                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1437                                         }
1438                                 }
1439                         }
1440                 }
1441
1442                 if (isset($data["notify"]) && isset($data["pubkey"])
1443                         && isset($data["poll"])
1444                         && isset($data["url"])
1445                 ) {
1446                         $data["network"] = Protocol::OSTATUS;
1447                 } elseif ($short) {
1448                         return false;
1449                 } else {
1450                         return $data;
1451                 }
1452
1453                 if ($short) {
1454                         return true;
1455                 }
1456
1457                 // Fetch all additional data from the feed
1458                 $curlResult = Network::curl($data["poll"]);
1459                 if ($curlResult->isTimeout()) {
1460                         self::$istimeout = true;
1461                         return $data;
1462                 }
1463                 $feed = $curlResult->getBody();
1464                 $dummy1 = null;
1465                 $dummy2 = null;
1466                 $dummy2 = null;
1467                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1468                 if (!$feed_data) {
1469                         return $data;
1470                 }
1471
1472                 if (!empty($feed_data["header"]["author-name"])) {
1473                         $data["name"] = $feed_data["header"]["author-name"];
1474                 }
1475                 if (!empty($feed_data["header"]["author-nick"])) {
1476                         $data["nick"] = $feed_data["header"]["author-nick"];
1477                 }
1478                 if (!empty($feed_data["header"]["author-avatar"])) {
1479                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1480                 }
1481                 if (!empty($feed_data["header"]["author-id"])) {
1482                         $data["alias"] = $feed_data["header"]["author-id"];
1483                 }
1484                 if (!empty($feed_data["header"]["author-location"])) {
1485                         $data["location"] = $feed_data["header"]["author-location"];
1486                 }
1487                 if (!empty($feed_data["header"]["author-about"])) {
1488                         $data["about"] = $feed_data["header"]["author-about"];
1489                 }
1490                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1491                 // So we take the value that we just fetched, although the other one worked as well
1492                 if (!empty($feed_data["header"]["author-link"])) {
1493                         $data["url"] = $feed_data["header"]["author-link"];
1494                 }
1495
1496                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1497                         $data['url'] = $data["alias"];
1498                         $data["alias"] = '';
1499                 }
1500
1501                 /// @todo Fetch location and "about" from the feed as well
1502                 return $data;
1503         }
1504
1505         /**
1506          * @brief Fetch data from a pump.io profile page
1507          *
1508          * @param string $profile_link Link to the profile page
1509          *
1510          * @return array profile data
1511          */
1512         private static function pumpioProfileData($profile_link)
1513         {
1514                 $doc = new DOMDocument();
1515                 if (!@$doc->loadHTMLFile($profile_link)) {
1516                         return false;
1517                 }
1518
1519                 $xpath = new DomXPath($doc);
1520
1521                 $data = [];
1522
1523                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1524
1525                 if ($data["name"] == '') {
1526                         // This is ugly - but pump.io doesn't seem to know a better way for it
1527                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1528                         $pos = strpos($data["name"], chr(10));
1529                         if ($pos) {
1530                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1531                         }
1532                 }
1533
1534                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1535
1536                 if ($data["location"] == '') {
1537                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1538                 }
1539
1540                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1541
1542                 if ($data["about"] == '') {
1543                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1544                 }
1545
1546                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1547                 if (!$avatar) {
1548                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1549                 }
1550                 if ($avatar) {
1551                         foreach ($avatar->attributes as $attribute) {
1552                                 if ($attribute->name == "src") {
1553                                         $data["photo"] = trim($attribute->value);
1554                                 }
1555                         }
1556                 }
1557
1558                 return $data;
1559         }
1560
1561         /**
1562          * @brief Check for pump.io contact
1563          *
1564          * @param array  $webfinger Webfinger data
1565          * @param string $addr
1566          * @return array pump.io data
1567          */
1568         private static function pumpio($webfinger, $addr)
1569         {
1570                 $data = [];
1571                 // The array is reversed to take into account the order of preference for same-rel links
1572                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1573                 foreach (array_reverse($webfinger["links"]) as $link) {
1574                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1575                                 && (($link["type"] ?? "") == "text/html")
1576                                 && ($link["href"] != "")
1577                         ) {
1578                                 $data["url"] = $link["href"];
1579                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1580                                 $data["notify"] = $link["href"];
1581                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1582                                 $data["poll"] = $link["href"];
1583                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1584                                 $data["dialback"] = $link["href"];
1585                         }
1586                 }
1587                 if (isset($data["poll"]) && isset($data["notify"])
1588                         && isset($data["dialback"])
1589                         && isset($data["url"])
1590                 ) {
1591                         // by now we use these fields only for the network type detection
1592                         // So we unset all data that isn't used at the moment
1593                         unset($data["dialback"]);
1594
1595                         $data["network"] = Protocol::PUMPIO;
1596                 } else {
1597                         return false;
1598                 }
1599
1600                 $profile_data = self::pumpioProfileData($data["url"]);
1601
1602                 if (!$profile_data) {
1603                         return false;
1604                 }
1605
1606                 $data = array_merge($data, $profile_data);
1607
1608                 if (($addr != '') && ($data['name'] != '')) {
1609                         $name = trim(str_replace($addr, '', $data['name']));
1610                         if ($name != '') {
1611                                 $data['name'] = $name;
1612                         }
1613                 }
1614
1615                 return $data;
1616         }
1617
1618         /**
1619          * @brief Check for twitter contact
1620          *
1621          * @param string $uri
1622          *
1623          * @return array twitter data
1624          */
1625         private static function twitter($uri)
1626         {
1627                 if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) {
1628                         $nick = $matches[1];
1629                 } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) {
1630                         $nick = $matches[1];
1631                 } else {
1632                         return [];
1633                 }
1634
1635                 $data = [];
1636                 $data['url'] = 'https://twitter.com/' . $nick;
1637                 $data['addr'] = $nick . '@twitter.com';
1638                 $data['nick'] = $data['name'] = $nick;
1639                 $data['network'] = Protocol::TWITTER;
1640                 $data['baseurl'] = 'https://twitter.com';
1641
1642                 $curlResult = Network::curl($data['url'], false);
1643                 if (!$curlResult->isSuccess()) {
1644                         return [];
1645                 }
1646
1647                 $body = $curlResult->getBody();
1648                 $doc = new DOMDocument();
1649                 @$doc->loadHTML($body);
1650                 $xpath = new DOMXPath($doc);
1651
1652                 $list = $xpath->query('//img[@class]');
1653                 foreach ($list as $node) {
1654                         $img_attr = [];
1655                         if ($node->attributes->length) {
1656                                 foreach ($node->attributes as $attribute) {
1657                                         $img_attr[$attribute->name] = $attribute->value;
1658                                 }
1659                         }
1660
1661                         if (empty($img_attr['class'])) {
1662                                 continue;
1663                         }
1664
1665                         if (strpos($img_attr['class'], 'ProfileAvatar-image') !== false) {
1666                                 if (!empty($img_attr['src'])) {
1667                                         $data['photo'] = $img_attr['src'];
1668                                 }
1669                                 if (!empty($img_attr['alt'])) {
1670                                         $data['name'] = $img_attr['alt'];
1671                                 }
1672                         }
1673                 }
1674
1675                 return $data;
1676         }
1677
1678         /**
1679          * @brief Check page for feed link
1680          *
1681          * @param string $url Page link
1682          *
1683          * @return string feed link
1684          */
1685         private static function getFeedLink($url)
1686         {
1687                 $doc = new DOMDocument();
1688
1689                 if (!@$doc->loadHTMLFile($url)) {
1690                         return false;
1691                 }
1692
1693                 $xpath = new DomXPath($doc);
1694
1695                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1696                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1697                 if (!is_object($feeds)) {
1698                         return false;
1699                 }
1700
1701                 if ($feeds->length == 0) {
1702                         return false;
1703                 }
1704
1705                 $feed_url = "";
1706
1707                 foreach ($feeds as $feed) {
1708                         $attr = [];
1709                         foreach ($feed->attributes as $attribute) {
1710                                 $attr[$attribute->name] = trim($attribute->value);
1711                         }
1712
1713                         if (empty($feed_url) && !empty($attr['href'])) {
1714                                 $feed_url = $attr["href"];
1715                         }
1716                 }
1717
1718                 return $feed_url;
1719         }
1720
1721         /**
1722          * @brief Check for feed contact
1723          *
1724          * @param string  $url   Profile link
1725          * @param boolean $probe Do a probe if the page contains a feed link
1726          *
1727          * @return array feed data
1728          * @throws HTTPException\InternalServerErrorException
1729          */
1730         private static function feed($url, $probe = true)
1731         {
1732                 $curlResult = Network::curl($url);
1733                 if ($curlResult->isTimeout()) {
1734                         self::$istimeout = true;
1735                         return false;
1736                 }
1737                 $feed = $curlResult->getBody();
1738                 $dummy1 = $dummy2 = $dummy3 = null;
1739                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1740
1741                 if (!$feed_data) {
1742                         if (!$probe) {
1743                                 return false;
1744                         }
1745
1746                         $feed_url = self::getFeedLink($url);
1747
1748                         if (!$feed_url) {
1749                                 return false;
1750                         }
1751
1752                         return self::feed($feed_url, false);
1753                 }
1754
1755                 if (!empty($feed_data["header"]["author-name"])) {
1756                         $data["name"] = $feed_data["header"]["author-name"];
1757                 }
1758
1759                 if (!empty($feed_data["header"]["author-nick"])) {
1760                         $data["nick"] = $feed_data["header"]["author-nick"];
1761                 }
1762
1763                 if (!empty($feed_data["header"]["author-avatar"])) {
1764                         $data["photo"] = $feed_data["header"]["author-avatar"];
1765                 }
1766
1767                 if (!empty($feed_data["header"]["author-id"])) {
1768                         $data["alias"] = $feed_data["header"]["author-id"];
1769                 }
1770
1771                 $data["url"] = $url;
1772                 $data["poll"] = $url;
1773
1774                 if (!empty($feed_data["header"]["author-link"])) {
1775                         $data["baseurl"] = $feed_data["header"]["author-link"];
1776                 } else {
1777                         $data["baseurl"] = $data["url"];
1778                 }
1779
1780                 $data["network"] = Protocol::FEED;
1781
1782                 return $data;
1783         }
1784
1785         /**
1786          * @brief Check for mail contact
1787          *
1788          * @param string  $uri Profile link
1789          * @param integer $uid User ID
1790          *
1791          * @return array mail data
1792          * @throws \Exception
1793          */
1794         private static function mail($uri, $uid)
1795         {
1796                 if (!Network::isEmailDomainValid($uri)) {
1797                         return false;
1798                 }
1799
1800                 if ($uid == 0) {
1801                         return false;
1802                 }
1803
1804                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1805
1806                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1807                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1808                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1809
1810                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1811                         return false;
1812                 }
1813
1814                 $mailbox = Email::constructMailboxName($mailacct);
1815                 $password = '';
1816                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1817                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1818                 if (!$mbox) {
1819                         return false;
1820                 }
1821
1822                 $msgs = Email::poll($mbox, $uri);
1823                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1824
1825                 if (!count($msgs)) {
1826                         return false;
1827                 }
1828
1829                 $phost = substr($uri, strpos($uri, '@') + 1);
1830
1831                 $data = [];
1832                 $data["addr"]    = $uri;
1833                 $data["network"] = Protocol::MAIL;
1834                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1835                 $data["nick"]    = $data["name"];
1836                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1837                 $data["url"]     = 'mailto:'.$uri;
1838                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1839                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1840
1841                 $x = Email::messageMeta($mbox, $msgs[0]);
1842                 if (stristr($x[0]->from, $uri)) {
1843                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1844                 } elseif (stristr($x[0]->to, $uri)) {
1845                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1846                 }
1847                 if (isset($adr)) {
1848                         foreach ($adr as $feadr) {
1849                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1850                                         &&(strcasecmp($feadr->host, $phost) == 0)
1851                                         && (strlen($feadr->personal))
1852                                 ) {
1853                                         $personal = imap_mime_header_decode($feadr->personal);
1854                                         $data["name"] = "";
1855                                         foreach ($personal as $perspart) {
1856                                                 if ($perspart->charset != "default") {
1857                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1858                                                 } else {
1859                                                         $data["name"] .= $perspart->text;
1860                                                 }
1861                                         }
1862
1863                                         $data["name"] = Strings::escapeTags($data["name"]);
1864                                 }
1865                         }
1866                 }
1867                 if (!empty($mbox)) {
1868                         imap_close($mbox);
1869                 }
1870                 return $data;
1871         }
1872
1873         /**
1874          * @brief Mix two paths together to possibly fix missing parts
1875          *
1876          * @param string $avatar Path to the avatar
1877          * @param string $base   Another path that is hopefully complete
1878          *
1879          * @return string fixed avatar path
1880          * @throws \Exception
1881          */
1882         public static function fixAvatar($avatar, $base)
1883         {
1884                 $base_parts = parse_url($base);
1885
1886                 // Remove all parts that could create a problem
1887                 unset($base_parts['path']);
1888                 unset($base_parts['query']);
1889                 unset($base_parts['fragment']);
1890
1891                 $avatar_parts = parse_url($avatar);
1892
1893                 // Now we mix them
1894                 $parts = array_merge($base_parts, $avatar_parts);
1895
1896                 // And put them together again
1897                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1898                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1899                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1900                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1901                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1902                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1903
1904                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1905
1906                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1907
1908                 return $fixed;
1909         }
1910 }