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