src/Network/Probe.php

   1 <?php
   2 /**
   3  * @file src/Network/Probe.php
   4  */
   5 namespace Friendica\Network;
   6
   7 /**
   8  * @file src/Network/Probe.php
   9  * @brief Functions for probing URL
  10  */
  11
  12 use DOMDocument;
  13 use DomXPath;
  14 use Friendica\Core\Cache;
  15 use Friendica\Core\Config;
  16 use Friendica\Core\Logger;
  17 use Friendica\Core\Protocol;
  18 use Friendica\Core\System;
  19 use Friendica\Database\DBA;
  20 use Friendica\Model\Profile;
  21 use Friendica\Protocol\Activity\Namespaces;
  22 use Friendica\Protocol\ActivityPub;
  23 use Friendica\Protocol\Email;
  24 use Friendica\Protocol\Feed;
  25 use Friendica\Util\Crypto;
  26 use Friendica\Util\DateTimeFormat;
  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 = [];
 147                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 148                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 149                 // compared to the XML webfinger. So this is commented out by now.
 150                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 151
 152                 foreach ($links["xrd"]["link"] as $value => $link) {
 153                         if (!empty($link["@attributes"])) {
 154                                 $attributes = $link["@attributes"];
 155                         } elseif ($value == "@attributes") {
 156                                 $attributes = $link;
 157                         } else {
 158                                 continue;
 159                         }
 160
 161                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 162                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 163
 164                                 $lrdd[$type] = $attributes["template"];
 165                         }
 166                 }
 167
 168                 self::$baseurl = $host_url;
 169
 170                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
 171
 172                 return $lrdd;
 173         }
 174
 175         /**
 176          * @brief Perform Webfinger lookup and return DFRN data
 177          *
 178          * Given an email style address, perform webfinger lookup and
 179          * return the resulting DFRN profile URL, or if no DFRN profile URL
 180          * is located, returns an OStatus subscription template (prefixed
 181          * with the string 'stat:' to identify it as on OStatus template).
 182          * If this isn't an email style address just return $webbie.
 183          * Return an empty string if email-style addresses but webfinger fails,
 184          * or if the resultant personal XRD doesn't contain a supported
 185          * subscription/friend-request attribute.
 186          *
 187          * amended 7/9/2011 to return an hcard which could save potentially loading
 188          * a lengthy content page to scrape dfrn attributes
 189          *
 190          * @param string $webbie    Address that should be probed
 191          * @param string $hcard_url Link to the hcard - is returned by reference
 192          *
 193          * @return string profile link
 194          * @throws HTTPException\InternalServerErrorException
 195          */
 196         public static function webfingerDfrn($webbie, &$hcard_url)
 197         {
 198                 $profile_link = '';
 199
 200                 $links = self::lrdd($webbie);
 201                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
 202                 if (!empty($links) && is_array($links)) {
 203                         foreach ($links as $link) {
 204                                 if ($link['@attributes']['rel'] === Namespaces::DFRN) {
 205                                         $profile_link = $link['@attributes']['href'];
 206                                 }
 207                                 if (($link['@attributes']['rel'] === Namespaces::OSTATUSSUB) && ($profile_link == "")) {
 208                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 209                                 }
 210                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 211                                         $hcard_url = $link['@attributes']['href'];
 212                                 }
 213                         }
 214                 }
 215                 return $profile_link;
 216         }
 217
 218         /**
 219          * @brief Check an URI for LRDD data
 220          *
 221          * this is a replacement for the "lrdd" function.
 222          * It isn't used in this class and has some redundancies in the code.
 223          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 224          *
 225          * @param string $uri Address that should be probed
 226          *
 227          * @return array uri data
 228          * @throws HTTPException\InternalServerErrorException
 229          */
 230         public static function lrdd($uri)
 231         {
 232                 $lrdd = self::hostMeta($uri);
 233                 $webfinger = null;
 234
 235                 if (is_bool($lrdd)) {
 236                         return [];
 237                 }
 238
 239                 if (!$lrdd) {
 240                         $parts = @parse_url($uri);
 241                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 242                                 return [];
 243                         }
 244
 245                         $host = $parts["host"];
 246                         if (!empty($parts["port"])) {
 247                                 $host .= ':'.$parts["port"];
 248                         }
 249
 250                         $path_parts = explode("/", trim($parts["path"], "/"));
 251
 252                         $nick = array_pop($path_parts);
 253
 254                         do {
 255                                 $lrdd = self::hostMeta($host);
 256                                 $host .= "/".array_shift($path_parts);
 257                         } while (!$lrdd && (sizeof($path_parts) > 0));
 258                 }
 259
 260                 if (!$lrdd) {
 261                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
 262                         return [];
 263                 }
 264
 265                 foreach ($lrdd as $type => $template) {
 266                         if ($webfinger) {
 267                                 continue;
 268                         }
 269
 270                         $path = str_replace('{uri}', urlencode($uri), $template);
 271                         $webfinger = self::webfinger($path, $type);
 272
 273                         if (!$webfinger && (strstr($uri, "@"))) {
 274                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 275                                 $webfinger = self::webfinger($path, $type);
 276                         }
 277
 278                         // Special treatment for Mastodon
 279                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 280                         // But the webfinger for this format fails.
 281                         if (!$webfinger && !empty($nick)) {
 282                                 // Mastodon uses a "@" as prefix for usernames in their url format
 283                                 $nick = ltrim($nick, '@');
 284
 285                                 $addr = $nick."@".$host;
 286
 287                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 288                                 $webfinger = self::webfinger($path, $type);
 289                         }
 290                 }
 291
 292                 if (!is_array($webfinger["links"])) {
 293                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
 294                         return false;
 295                 }
 296
 297                 $data = [];
 298
 299                 foreach ($webfinger["links"] as $link) {
 300                         $data[] = ["@attributes" => $link];
 301                 }
 302
 303                 if (is_array($webfinger["aliases"])) {
 304                         foreach ($webfinger["aliases"] as $alias) {
 305                                 $data[] = ["@attributes" =>
 306                                                         ["rel" => "alias",
 307                                                                 "href" => $alias]];
 308                         }
 309                 }
 310
 311                 return $data;
 312         }
 313
 314         /**
 315          * @brief Fetch information (protocol endpoints and user information) about a given uri
 316          *
 317          * @param string  $uri     Address that should be probed
 318          * @param string  $network Test for this specific network
 319          * @param integer $uid     User ID for the probe (only used for mails)
 320          * @param boolean $cache   Use cached values?
 321          *
 322          * @return array uri data
 323          * @throws HTTPException\InternalServerErrorException
 324          * @throws \ImagickException
 325          */
 326         public static function uri($uri, $network = '', $uid = -1, $cache = true)
 327         {
 328                 if ($cache) {
 329                         $result = Cache::get('Probe::uri:' . $network . ':' . $uri);
 330                         if (!is_null($result)) {
 331                                 return $result;
 332                         }
 333                 }
 334
 335                 if ($uid == -1) {
 336                         $uid = local_user();
 337                 }
 338
 339                 self::$istimeout = false;
 340
 341                 if ($network != Protocol::ACTIVITYPUB) {
 342                         $data = self::detect($uri, $network, $uid);
 343                 } else {
 344                         $data = null;
 345                 }
 346
 347                 // When the previous detection process had got a time out
 348                 // we could falsely detect a Friendica profile as AP profile.
 349                 if (!self::$istimeout) {
 350                         $ap_profile = ActivityPub::probeProfile($uri);
 351
 352                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 353                                 $data = $ap_profile;
 354                         } elseif (!empty($ap_profile)) {
 355                                 $ap_profile['batch'] = '';
 356                                 $data = array_merge($ap_profile, $data);
 357                         }
 358                 } else {
 359                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 360                 }
 361
 362                 if (!isset($data['url'])) {
 363                         $data['url'] = $uri;
 364                 }
 365
 366                 if (!empty($data['photo'])) {
 367                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl'] ?? ''), Strings::normaliseLink($data['photo']));
 368                 } else {
 369                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 370                 }
 371
 372                 if (empty($data['name'])) {
 373                         if (!empty($data['nick'])) {
 374                                 $data['name'] = $data['nick'];
 375                         }
 376
 377                         if (empty($data['name'])) {
 378                                 $data['name'] = $data['url'];
 379                         }
 380                 }
 381
 382                 if (empty($data['nick'])) {
 383                         $data['nick'] = strtolower($data['name']);
 384
 385                         if (strpos($data['nick'], ' ')) {
 386                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 387                         }
 388                 }
 389
 390                 if (!empty(self::$baseurl)) {
 391                         $data['baseurl'] = self::$baseurl;
 392                 }
 393
 394                 if (empty($data['network'])) {
 395                         $data['network'] = Protocol::PHANTOM;
 396                 }
 397
 398                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 399                         $data['hide'] = self::getHideStatus($data['url']);
 400                 }
 401
 402                 $data = self::rearrangeData($data);
 403
 404                 // Only store into the cache if the value seems to be valid
 405                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 406                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 407                 }
 408
 409                 return $data;
 410         }
 411
 412
 413         /**
 414          * Fetches the "hide" status from the profile
 415          *
 416          * @param string $url URL of the profile
 417          *
 418          * @return boolean "hide" status
 419          */
 420         private static function getHideStatus($url)
 421         {
 422                 $curlResult = Network::curl($url);
 423                 if (!$curlResult->isSuccess()) {
 424                         return false;
 425                 }
 426
 427                 // If the file is too large then exit
 428                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 429                         return false;
 430                 }
 431
 432                 // If it isn't a HTML file then exit
 433                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 434                         return false;
 435                 }
 436
 437                 $body = $curlResult->getBody();
 438
 439                 $doc = new DOMDocument();
 440                 @$doc->loadHTML($body);
 441
 442                 $xpath = new DOMXPath($doc);
 443
 444                 $list = $xpath->query('//meta[@name]');
 445                 foreach ($list as $node) {
 446                         $meta_tag = [];
 447                         if ($node->attributes->length) {
 448                                 foreach ($node->attributes as $attribute) {
 449                                         $meta_tag[$attribute->name] = $attribute->value;
 450                                 }
 451                         }
 452
 453                         if (empty($meta_tag['content'])) {
 454                                 continue;
 455                         }
 456
 457                         $content = strtolower(trim($meta_tag['content']));
 458
 459                         switch (strtolower(trim($meta_tag['name']))) {
 460                                 case 'dfrn-global-visibility':
 461                                         if ($content == 'false') {
 462                                                 return true;
 463                                         }
 464                                         break;
 465                                 case 'robots':
 466                                         if (strpos($content, 'noindex') !== false) {
 467                                                 return true;
 468                                         }
 469                                         break;
 470                         }
 471                 }
 472
 473                 return false;
 474         }
 475
 476         /**
 477          * @brief Checks if a profile url should be OStatus but only provides partial information
 478          *
 479          * @param array  $webfinger Webfinger data
 480          * @param string $lrdd      Path template for webfinger request
 481          * @param string $type      type
 482          *
 483          * @return array fixed webfinger data
 484          * @throws HTTPException\InternalServerErrorException
 485          */
 486         private static function fixOStatus($webfinger, $lrdd, $type)
 487         {
 488                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 489                         return $webfinger;
 490                 }
 491
 492                 $is_ostatus = false;
 493                 $has_key = false;
 494
 495                 foreach ($webfinger['links'] as $link) {
 496                         if ($link['rel'] == Namespaces::OSTATUSSUB) {
 497                                 $is_ostatus = true;
 498                         }
 499                         if ($link['rel'] == 'magic-public-key') {
 500                                 $has_key = true;
 501                         }
 502                 }
 503
 504                 if (!$is_ostatus || $has_key) {
 505                         return $webfinger;
 506                 }
 507
 508                 $url = Network::switchScheme($webfinger['subject']);
 509                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 510                 $webfinger2 = self::webfinger($path, $type);
 511
 512                 // Is the new webfinger detectable as OStatus?
 513                 if (self::ostatus($webfinger2, true)) {
 514                         $webfinger = $webfinger2;
 515                 }
 516
 517                 return $webfinger;
 518         }
 519
 520         /**
 521          * @brief Fetch information (protocol endpoints and user information) about a given uri
 522          *
 523          * This function is only called by the "uri" function that adds caching and rearranging of data.
 524          *
 525          * @param string  $uri     Address that should be probed
 526          * @param string  $network Test for this specific network
 527          * @param integer $uid     User ID for the probe (only used for mails)
 528          *
 529          * @return array uri data
 530          * @throws HTTPException\InternalServerErrorException
 531          */
 532         private static function detect($uri, $network, $uid)
 533         {
 534                 $parts = parse_url($uri);
 535
 536                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 537                         $host = $parts["host"];
 538                         if (!empty($parts["port"])) {
 539                                 $host .= ':'.$parts["port"];
 540                         }
 541
 542                         if ($host == 'twitter.com') {
 543                                 return self::twitter($uri);
 544                         }
 545                         $lrdd = self::hostMeta($host);
 546
 547                         if (is_bool($lrdd)) {
 548                                 return [];
 549                         }
 550
 551                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
 552
 553                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 554                                 $host .= "/".array_shift($path_parts);
 555                                 $lrdd = self::hostMeta($host);
 556                         }
 557                         if (!$lrdd) {
 558                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 559                                 return self::feed($uri);
 560                         }
 561                         $nick = array_pop($path_parts);
 562
 563                         // Mastodon uses a "@" as prefix for usernames in their url format
 564                         $nick = ltrim($nick, '@');
 565
 566                         $addr = $nick."@".$host;
 567                 } elseif (strstr($uri, '@')) {
 568                         // If the URI starts with "mailto:" then jump directly to the mail detection
 569                         if (strpos($uri, 'mailto:') !== false) {
 570                                 $uri = str_replace('mailto:', '', $uri);
 571                                 return self::mail($uri, $uid);
 572                         }
 573
 574                         if ($network == Protocol::MAIL) {
 575                                 return self::mail($uri, $uid);
 576                         }
 577                         // Remove "acct:" from the URI
 578                         $uri = str_replace('acct:', '', $uri);
 579
 580                         $host = substr($uri, strpos($uri, '@') + 1);
 581                         $nick = substr($uri, 0, strpos($uri, '@'));
 582
 583                         if (strpos($uri, '@twitter.com')) {
 584                                 return self::twitter($uri);
 585                         }
 586                         $lrdd = self::hostMeta($host);
 587
 588                         if (is_bool($lrdd)) {
 589                                 return [];
 590                         }
 591
 592                         if (!$lrdd) {
 593                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 594                                 return self::mail($uri, $uid);
 595                         }
 596                         $addr = $uri;
 597                 } else {
 598                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 599                         return false;
 600                 }
 601
 602                 $webfinger = false;
 603
 604                 /// @todo Do we need the prefix "acct:" or "acct://"?
 605
 606                 foreach ($lrdd as $type => $template) {
 607                         if ($webfinger) {
 608                                 continue;
 609                         }
 610
 611                         // At first try it with the given uri
 612                         $path = str_replace('{uri}', urlencode($uri), $template);
 613                         $webfinger = self::webfinger($path, $type);
 614
 615                         // Fix possible problems with GNU Social probing to wrong scheme
 616                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 617
 618                         // We cannot be sure that the detected address was correct, so we don't use the values
 619                         if ($webfinger && ($uri != $addr)) {
 620                                 $nick = "";
 621                                 $addr = "";
 622                         }
 623
 624                         // Try webfinger with the address (user@domain.tld)
 625                         if (!$webfinger) {
 626                                 $path = str_replace('{uri}', urlencode($addr), $template);
 627                                 $webfinger = self::webfinger($path, $type);
 628                         }
 629
 630                         // Mastodon needs to have it with "acct:"
 631                         if (!$webfinger) {
 632                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 633                                 $webfinger = self::webfinger($path, $type);
 634                         }
 635                 }
 636
 637                 if (!$webfinger) {
 638                         return self::feed($uri);
 639                 }
 640
 641                 $result = false;
 642
 643                 Logger::log("Probing ".$uri, Logger::DEBUG);
 644
 645                 if (in_array($network, ["", Protocol::DFRN])) {
 646                         $result = self::dfrn($webfinger);
 647                 }
 648                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 649                         $result = self::diaspora($webfinger);
 650                 }
 651                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 652                         $result = self::ostatus($webfinger);
 653                 }
 654                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 655                         $result = self::pumpio($webfinger, $addr);
 656                 }
 657                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 658                         $result = self::feed($uri);
 659                 } else {
 660                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 661                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 662                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 663                                 $result["nick"] = $nick;
 664                         }
 665
 666                         if (empty($result["addr"]) && ($addr != "")) {
 667                                 $result["addr"] = $addr;
 668                         }
 669                 }
 670
 671                 if (empty($result["network"])) {
 672                         $result["network"] = Protocol::PHANTOM;
 673                 }
 674
 675                 if (empty($result["url"])) {
 676                         $result["url"] = $uri;
 677                 }
 678
 679                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 680
 681                 if (empty($result["baseurl"])) {
 682                         $pos = strpos($result["url"], $host);
 683                         if ($pos) {
 684                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 685                         }
 686                 }
 687                 return $result;
 688         }
 689
 690         /**
 691          * @brief Perform a webfinger request.
 692          *
 693          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 694          *
 695          * @param string $url  Address that should be probed
 696          * @param string $type type
 697          *
 698          * @return array webfinger data
 699          * @throws HTTPException\InternalServerErrorException
 700          */
 701         private static function webfinger($url, $type)
 702         {
 703                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 704
 705                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 706                 if ($curlResult->isTimeout()) {
 707                         self::$istimeout = true;
 708                         return false;
 709                 }
 710                 $data = $curlResult->getBody();
 711
 712                 $webfinger = json_decode($data, true);
 713                 if (is_array($webfinger)) {
 714                         if (!isset($webfinger["links"])) {
 715                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 716                                 return false;
 717                         }
 718                         return $webfinger;
 719                 }
 720
 721                 // If it is not JSON, maybe it is XML
 722                 $xrd = XML::parseString($data, false);
 723                 if (!is_object($xrd)) {
 724                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 725                         return false;
 726                 }
 727
 728                 $xrd_arr = XML::elementToArray($xrd);
 729                 if (!isset($xrd_arr["xrd"]["link"])) {
 730                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 731                         return false;
 732                 }
 733
 734                 $webfinger = [];
 735
 736                 if (!empty($xrd_arr["xrd"]["subject"])) {
 737                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 738                 }
 739
 740                 if (!empty($xrd_arr["xrd"]["alias"])) {
 741                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 742                 }
 743
 744                 $webfinger["links"] = [];
 745
 746                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 747                         if (!empty($data["@attributes"])) {
 748                                 $attributes = $data["@attributes"];
 749                         } elseif ($value == "@attributes") {
 750                                 $attributes = $data;
 751                         } else {
 752                                 continue;
 753                         }
 754
 755                         $webfinger["links"][] = $attributes;
 756                 }
 757                 return $webfinger;
 758         }
 759
 760         /**
 761          * @brief Poll the Friendica specific noscrape page.
 762          *
 763          * "noscrape" is a faster alternative to fetch the data from the hcard.
 764          * This functionality was originally created for the directory.
 765          *
 766          * @param string $noscrape_url Link to the noscrape page
 767          * @param array  $data         The already fetched data
 768          *
 769          * @return array noscrape data
 770          * @throws HTTPException\InternalServerErrorException
 771          */
 772         private static function pollNoscrape($noscrape_url, $data)
 773         {
 774                 $curlResult = Network::curl($noscrape_url);
 775                 if ($curlResult->isTimeout()) {
 776                         self::$istimeout = true;
 777                         return false;
 778                 }
 779                 $content = $curlResult->getBody();
 780                 if (!$content) {
 781                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 782                         return false;
 783                 }
 784
 785                 $json = json_decode($content, true);
 786                 if (!is_array($json)) {
 787                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 788                         return false;
 789                 }
 790
 791                 if (!empty($json["fn"])) {
 792                         $data["name"] = $json["fn"];
 793                 }
 794
 795                 if (!empty($json["addr"])) {
 796                         $data["addr"] = $json["addr"];
 797                 }
 798
 799                 if (!empty($json["nick"])) {
 800                         $data["nick"] = $json["nick"];
 801                 }
 802
 803                 if (!empty($json["guid"])) {
 804                         $data["guid"] = $json["guid"];
 805                 }
 806
 807                 if (!empty($json["comm"])) {
 808                         $data["community"] = $json["comm"];
 809                 }
 810
 811                 if (!empty($json["tags"])) {
 812                         $keywords = implode(", ", $json["tags"]);
 813                         if ($keywords != "") {
 814                                 $data["keywords"] = $keywords;
 815                         }
 816                 }
 817
 818                 $location = Profile::formatLocation($json);
 819                 if ($location) {
 820                         $data["location"] = $location;
 821                 }
 822
 823                 if (!empty($json["about"])) {
 824                         $data["about"] = $json["about"];
 825                 }
 826
 827                 if (!empty($json["gender"])) {
 828                         $data["gender"] = $json["gender"];
 829                 }
 830
 831                 if (!empty($json["key"])) {
 832                         $data["pubkey"] = $json["key"];
 833                 }
 834
 835                 if (!empty($json["photo"])) {
 836                         $data["photo"] = $json["photo"];
 837                 }
 838
 839                 if (!empty($json["dfrn-request"])) {
 840                         $data["request"] = $json["dfrn-request"];
 841                 }
 842
 843                 if (!empty($json["dfrn-confirm"])) {
 844                         $data["confirm"] = $json["dfrn-confirm"];
 845                 }
 846
 847                 if (!empty($json["dfrn-notify"])) {
 848                         $data["notify"] = $json["dfrn-notify"];
 849                 }
 850
 851                 if (!empty($json["dfrn-poll"])) {
 852                         $data["poll"] = $json["dfrn-poll"];
 853                 }
 854
 855                 if (isset($json["hide"])) {
 856                         $data["hide"] = (bool)$json["hide"];
 857                 } else {
 858                         $data["hide"] = false;
 859                 }
 860
 861                 return $data;
 862         }
 863
 864         /**
 865          * @brief Check for valid DFRN data
 866          *
 867          * @param array $data DFRN data
 868          *
 869          * @return int Number of errors
 870          */
 871         public static function validDfrn($data)
 872         {
 873                 $errors = 0;
 874                 if (!isset($data['key'])) {
 875                         $errors ++;
 876                 }
 877                 if (!isset($data['dfrn-request'])) {
 878                         $errors ++;
 879                 }
 880                 if (!isset($data['dfrn-confirm'])) {
 881                         $errors ++;
 882                 }
 883                 if (!isset($data['dfrn-notify'])) {
 884                         $errors ++;
 885                 }
 886                 if (!isset($data['dfrn-poll'])) {
 887                         $errors ++;
 888                 }
 889                 return $errors;
 890         }
 891
 892         /**
 893          * @brief Fetch data from a DFRN profile page and via "noscrape"
 894          *
 895          * @param string $profile_link Link to the profile page
 896          *
 897          * @return array profile data
 898          * @throws HTTPException\InternalServerErrorException
 899          * @throws \ImagickException
 900          */
 901         public static function profile($profile_link)
 902         {
 903                 $data = [];
 904
 905                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
 906
 907                 // Fetch data via noscrape - this is faster
 908                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 909                 $data = self::pollNoscrape($noscrape_url, $data);
 910
 911                 if (!isset($data["notify"])
 912                         || !isset($data["confirm"])
 913                         || !isset($data["request"])
 914                         || !isset($data["poll"])
 915                         || !isset($data["name"])
 916                         || !isset($data["photo"])
 917                 ) {
 918                         $data = self::pollHcard($profile_link, $data, true);
 919                 }
 920
 921                 $prof_data = [];
 922
 923                 if (empty($data["addr"]) || empty($data["nick"])) {
 924                         $probe_data = self::uri($profile_link);
 925                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
 926                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
 927                 }
 928
 929                 $prof_data["addr"]         = $data["addr"];
 930                 $prof_data["nick"]         = $data["nick"];
 931                 $prof_data["dfrn-request"] = $data['request'] ?? null;
 932                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
 933                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
 934                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
 935                 $prof_data["photo"]        = $data['photo']   ?? null;
 936                 $prof_data["fn"]           = $data['name']    ?? null;
 937                 $prof_data["key"]          = $data['pubkey']  ?? null;
 938
 939                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
 940
 941                 return $prof_data;
 942         }
 943
 944         /**
 945          * @brief Check for DFRN contact
 946          *
 947          * @param array $webfinger Webfinger data
 948          *
 949          * @return array DFRN data
 950          * @throws HTTPException\InternalServerErrorException
 951          */
 952         private static function dfrn($webfinger)
 953         {
 954                 $hcard_url = "";
 955                 $data = [];
 956                 // The array is reversed to take into account the order of preference for same-rel links
 957                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
 958                 foreach (array_reverse($webfinger["links"]) as $link) {
 959                         if (($link["rel"] == Namespaces::DFRN) && !empty($link["href"])) {
 960                                 $data["network"] = Protocol::DFRN;
 961                         } elseif (($link["rel"] == Namespaces::FEED) && !empty($link["href"])) {
 962                                 $data["poll"] = $link["href"];
 963                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
 964                                 $data["url"] = $link["href"];
 965                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
 966                                 $hcard_url = $link["href"];
 967                         } elseif (($link["rel"] == Namespaces::POCO) && !empty($link["href"])) {
 968                                 $data["poco"] = $link["href"];
 969                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
 970                                 $data["photo"] = $link["href"];
 971                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
 972                                 $data["baseurl"] = trim($link["href"], '/');
 973                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
 974                                 $data["guid"] = $link["href"];
 975                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
 976                                 $data["pubkey"] = base64_decode($link["href"]);
 977
 978                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 979                                 if (strstr($data["pubkey"], 'RSA ')) {
 980                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 981                                 }
 982                         }
 983                 }
 984
 985                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 986                         foreach ($webfinger["aliases"] as $alias) {
 987                                 if (empty($data["url"]) && !strstr($alias, "@")) {
 988                                         $data["url"] = $alias;
 989                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
 990                                         $data["alias"] = $alias;
 991                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 992                                         $data["addr"] = substr($alias, 5);
 993                                 }
 994                         }
 995                 }
 996
 997                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 998                         $data["addr"] = substr($webfinger["subject"], 5);
 999                 }
1000
1001                 if (!isset($data["network"]) || ($hcard_url == "")) {
1002                         return false;
1003                 }
1004
1005                 // Fetch data via noscrape - this is faster
1006                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1007                 $data = self::pollNoscrape($noscrape_url, $data);
1008
1009                 if (isset($data["notify"])
1010                         && isset($data["confirm"])
1011                         && isset($data["request"])
1012                         && isset($data["poll"])
1013                         && isset($data["name"])
1014                         && isset($data["photo"])
1015                 ) {
1016                         return $data;
1017                 }
1018
1019                 $data = self::pollHcard($hcard_url, $data, true);
1020
1021                 return $data;
1022         }
1023
1024         /**
1025          * @brief Poll the hcard page (Diaspora and Friendica specific)
1026          *
1027          * @param string  $hcard_url Link to the hcard page
1028          * @param array   $data      The already fetched data
1029          * @param boolean $dfrn      Poll DFRN specific data
1030          *
1031          * @return array hcard data
1032          * @throws HTTPException\InternalServerErrorException
1033          */
1034         private static function pollHcard($hcard_url, $data, $dfrn = false)
1035         {
1036                 $curlResult = Network::curl($hcard_url);
1037                 if ($curlResult->isTimeout()) {
1038                         self::$istimeout = true;
1039                         return false;
1040                 }
1041                 $content = $curlResult->getBody();
1042                 if (!$content) {
1043                         return false;
1044                 }
1045
1046                 $doc = new DOMDocument();
1047                 if (!@$doc->loadHTML($content)) {
1048                         return false;
1049                 }
1050
1051                 $xpath = new DomXPath($doc);
1052
1053                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1054                 if (!is_object($vcards)) {
1055                         return false;
1056                 }
1057
1058                 if (!isset($data["baseurl"])) {
1059                         $data["baseurl"] = "";
1060                 }
1061
1062                 if ($vcards->length > 0) {
1063                         $vcard = $vcards->item(0);
1064
1065                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1066                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1067                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1068                         if (($search->length > 0) && empty($data["guid"])) {
1069                                 $data["guid"] = $search->item(0)->nodeValue;
1070                         }
1071
1072                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1073                         if ($search->length > 0) {
1074                                 $data["nick"] = $search->item(0)->nodeValue;
1075                         }
1076
1077                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1078                         if ($search->length > 0) {
1079                                 $data["name"] = $search->item(0)->nodeValue;
1080                         }
1081
1082                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1083                         if ($search->length > 0) {
1084                                 $data["searchable"] = $search->item(0)->nodeValue;
1085                         }
1086
1087                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1088                         if ($search->length > 0) {
1089                                 $data["pubkey"] = $search->item(0)->nodeValue;
1090                                 if (strstr($data["pubkey"], 'RSA ')) {
1091                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1092                                 }
1093                         }
1094
1095                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1096                         if ($search->length > 0) {
1097                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1098                         }
1099                 }
1100
1101                 $avatar = [];
1102                 if (!empty($vcard)) {
1103                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1104                         foreach ($photos as $photo) {
1105                                 $attr = [];
1106                                 foreach ($photo->attributes as $attribute) {
1107                                         $attr[$attribute->name] = trim($attribute->value);
1108                                 }
1109
1110                                 if (isset($attr["src"]) && isset($attr["width"])) {
1111                                         $avatar[$attr["width"]] = $attr["src"];
1112                                 }
1113
1114                                 // We don't have a width. So we just take everything that we got.
1115                                 // This is a Hubzilla workaround which doesn't send a width.
1116                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1117                                         $avatar[] = $attr["src"];
1118                                 }
1119                         }
1120                 }
1121
1122                 if (sizeof($avatar)) {
1123                         ksort($avatar);
1124                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1125                 }
1126
1127                 if ($dfrn) {
1128                         // Poll DFRN specific data
1129                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1130                         if ($search->length > 0) {
1131                                 foreach ($search as $link) {
1132                                         //$data["request"] = $search->item(0)->nodeValue;
1133                                         $attr = [];
1134                                         foreach ($link->attributes as $attribute) {
1135                                                 $attr[$attribute->name] = trim($attribute->value);
1136                                         }
1137
1138                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1139                                 }
1140                         }
1141
1142                         // Older Friendica versions had used the "uid" field differently than newer versions
1143                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1144                                 unset($data["guid"]);
1145                         }
1146                 }
1147
1148
1149                 return $data;
1150         }
1151
1152         /**
1153          * @brief Check for Diaspora contact
1154          *
1155          * @param array $webfinger Webfinger data
1156          *
1157          * @return array Diaspora data
1158          * @throws HTTPException\InternalServerErrorException
1159          */
1160         private static function diaspora($webfinger)
1161         {
1162                 $hcard_url = "";
1163                 $data = [];
1164                 // The array is reversed to take into account the order of preference for same-rel links
1165                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1166                 foreach (array_reverse($webfinger["links"]) as $link) {
1167                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1168                                 $hcard_url = $link["href"];
1169                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1170                                 $data["baseurl"] = trim($link["href"], '/');
1171                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1172                                 $data["guid"] = $link["href"];
1173                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1174                                 $data["url"] = $link["href"];
1175                         } elseif (($link["rel"] == Namespaces::FEED) && !empty($link["href"])) {
1176                                 $data["poll"] = $link["href"];
1177                         } elseif (($link["rel"] == Namespaces::POCO) && !empty($link["href"])) {
1178                                 $data["poco"] = $link["href"];
1179                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1180                                 $data["notify"] = $link["href"];
1181                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1182                                 $data["pubkey"] = base64_decode($link["href"]);
1183
1184                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1185                                 if (strstr($data["pubkey"], 'RSA ')) {
1186                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1187                                 }
1188                         }
1189                 }
1190
1191                 if (!isset($data["url"]) || ($hcard_url == "")) {
1192                         return false;
1193                 }
1194
1195                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1196                         foreach ($webfinger["aliases"] as $alias) {
1197                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1198                                         $data["alias"] = $alias;
1199                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1200                                         $data["addr"] = substr($alias, 5);
1201                                 }
1202                         }
1203                 }
1204
1205                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1206                         $data["addr"] = substr($webfinger["subject"], 5);
1207                 }
1208
1209                 // Fetch further information from the hcard
1210                 $data = self::pollHcard($hcard_url, $data);
1211
1212                 if (!$data) {
1213                         return false;
1214                 }
1215
1216                 if (isset($data["url"])
1217                         && isset($data["guid"])
1218                         && isset($data["baseurl"])
1219                         && isset($data["pubkey"])
1220                         && ($hcard_url != "")
1221                 ) {
1222                         $data["network"] = Protocol::DIASPORA;
1223
1224                         // The Diaspora handle must always be lowercase
1225                         if (!empty($data["addr"])) {
1226                                 $data["addr"] = strtolower($data["addr"]);
1227                         }
1228
1229                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1230                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1231                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1232                 } else {
1233                         return false;
1234                 }
1235
1236                 return $data;
1237         }
1238
1239         /**
1240          * @brief Check for OStatus contact
1241          *
1242          * @param array $webfinger Webfinger data
1243          * @param bool  $short     Short detection mode
1244          *
1245          * @return array|bool OStatus data or "false" on error or "true" on short mode
1246          * @throws HTTPException\InternalServerErrorException
1247          */
1248         private static function ostatus($webfinger, $short = false)
1249         {
1250                 $data = [];
1251
1252                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1253                         foreach ($webfinger["aliases"] as $alias) {
1254                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1255                                         $data["addr"] = str_replace('acct:', '', $alias);
1256                                 }
1257                         }
1258                 }
1259
1260                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1261                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1262                 ) {
1263                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1264                 }
1265
1266                 if (is_array($webfinger["links"])) {
1267                         // The array is reversed to take into account the order of preference for same-rel links
1268                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1269                         foreach (array_reverse($webfinger["links"]) as $link) {
1270                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1271                                         && (($link["type"] ?? "") == "text/html")
1272                                         && ($link["href"] != "")
1273                                 ) {
1274                                         $data["url"] = $link["href"];
1275                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1276                                         $data["notify"] = $link["href"];
1277                                 } elseif (($link["rel"] == Namespaces::FEED) && !empty($link["href"])) {
1278                                         $data["poll"] = $link["href"];
1279                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1280                                         $pubkey = $link["href"];
1281
1282                                         if (substr($pubkey, 0, 5) === 'data:') {
1283                                                 if (strstr($pubkey, ',')) {
1284                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1285                                                 } else {
1286                                                         $pubkey = substr($pubkey, 5);
1287                                                 }
1288                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1289                                                 $curlResult = Network::curl($pubkey);
1290                                                 if ($curlResult->isTimeout()) {
1291                                                         self::$istimeout = true;
1292                                                         return false;
1293                                                 }
1294                                                 $pubkey = $curlResult->getBody();
1295                                         }
1296
1297                                         $key = explode(".", $pubkey);
1298
1299                                         if (sizeof($key) >= 3) {
1300                                                 $m = Strings::base64UrlDecode($key[1]);
1301                                                 $e = Strings::base64UrlDecode($key[2]);
1302                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1303                                         }
1304                                 }
1305                         }
1306                 }
1307
1308                 if (isset($data["notify"]) && isset($data["pubkey"])
1309                         && isset($data["poll"])
1310                         && isset($data["url"])
1311                 ) {
1312                         $data["network"] = Protocol::OSTATUS;
1313                 } else {
1314                         return false;
1315                 }
1316
1317                 if ($short) {
1318                         return true;
1319                 }
1320
1321                 // Fetch all additional data from the feed
1322                 $curlResult = Network::curl($data["poll"]);
1323                 if ($curlResult->isTimeout()) {
1324                         self::$istimeout = true;
1325                         return false;
1326                 }
1327                 $feed = $curlResult->getBody();
1328                 $dummy1 = null;
1329                 $dummy2 = null;
1330                 $dummy2 = null;
1331                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1332                 if (!$feed_data) {
1333                         return false;
1334                 }
1335
1336                 if (!empty($feed_data["header"]["author-name"])) {
1337                         $data["name"] = $feed_data["header"]["author-name"];
1338                 }
1339                 if (!empty($feed_data["header"]["author-nick"])) {
1340                         $data["nick"] = $feed_data["header"]["author-nick"];
1341                 }
1342                 if (!empty($feed_data["header"]["author-avatar"])) {
1343                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1344                 }
1345                 if (!empty($feed_data["header"]["author-id"])) {
1346                         $data["alias"] = $feed_data["header"]["author-id"];
1347                 }
1348                 if (!empty($feed_data["header"]["author-location"])) {
1349                         $data["location"] = $feed_data["header"]["author-location"];
1350                 }
1351                 if (!empty($feed_data["header"]["author-about"])) {
1352                         $data["about"] = $feed_data["header"]["author-about"];
1353                 }
1354                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1355                 // So we take the value that we just fetched, although the other one worked as well
1356                 if (!empty($feed_data["header"]["author-link"])) {
1357                         $data["url"] = $feed_data["header"]["author-link"];
1358                 }
1359
1360                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1361                         $data['url'] = $data["alias"];
1362                         $data["alias"] = '';
1363                 }
1364
1365                 /// @todo Fetch location and "about" from the feed as well
1366                 return $data;
1367         }
1368
1369         /**
1370          * @brief Fetch data from a pump.io profile page
1371          *
1372          * @param string $profile_link Link to the profile page
1373          *
1374          * @return array profile data
1375          */
1376         private static function pumpioProfileData($profile_link)
1377         {
1378                 $doc = new DOMDocument();
1379                 if (!@$doc->loadHTMLFile($profile_link)) {
1380                         return false;
1381                 }
1382
1383                 $xpath = new DomXPath($doc);
1384
1385                 $data = [];
1386
1387                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1388
1389                 if ($data["name"] == '') {
1390                         // This is ugly - but pump.io doesn't seem to know a better way for it
1391                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1392                         $pos = strpos($data["name"], chr(10));
1393                         if ($pos) {
1394                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1395                         }
1396                 }
1397
1398                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1399
1400                 if ($data["location"] == '') {
1401                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1402                 }
1403
1404                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1405
1406                 if ($data["about"] == '') {
1407                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1408                 }
1409
1410                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1411                 if (!$avatar) {
1412                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1413                 }
1414                 if ($avatar) {
1415                         foreach ($avatar->attributes as $attribute) {
1416                                 if ($attribute->name == "src") {
1417                                         $data["photo"] = trim($attribute->value);
1418                                 }
1419                         }
1420                 }
1421
1422                 return $data;
1423         }
1424
1425         /**
1426          * @brief Check for pump.io contact
1427          *
1428          * @param array $webfinger Webfinger data
1429          *
1430          * @param       $addr
1431          * @return array pump.io data
1432          */
1433         private static function pumpio($webfinger, $addr)
1434         {
1435                 $data = [];
1436                 // The array is reversed to take into account the order of preference for same-rel links
1437                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1438                 foreach (array_reverse($webfinger["links"]) as $link) {
1439                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1440                                 && (($link["type"] ?? "") == "text/html")
1441                                 && ($link["href"] != "")
1442                         ) {
1443                                 $data["url"] = $link["href"];
1444                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1445                                 $data["notify"] = $link["href"];
1446                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1447                                 $data["poll"] = $link["href"];
1448                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1449                                 $data["dialback"] = $link["href"];
1450                         }
1451                 }
1452                 if (isset($data["poll"]) && isset($data["notify"])
1453                         && isset($data["dialback"])
1454                         && isset($data["url"])
1455                 ) {
1456                         // by now we use these fields only for the network type detection
1457                         // So we unset all data that isn't used at the moment
1458                         unset($data["dialback"]);
1459
1460                         $data["network"] = Protocol::PUMPIO;
1461                 } else {
1462                         return false;
1463                 }
1464
1465                 $profile_data = self::pumpioProfileData($data["url"]);
1466
1467                 if (!$profile_data) {
1468                         return false;
1469                 }
1470
1471                 $data = array_merge($data, $profile_data);
1472
1473                 if (($addr != '') && ($data['name'] != '')) {
1474                         $name = trim(str_replace($addr, '', $data['name']));
1475                         if ($name != '') {
1476                                 $data['name'] = $name;
1477                         }
1478                 }
1479
1480                 return $data;
1481         }
1482
1483         /**
1484          * @brief Check for twitter contact
1485          *
1486          * @param string $uri
1487          *
1488          * @return array twitter data
1489          */
1490         private static function twitter($uri)
1491         {
1492                 if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) {
1493                         $nick = $matches[1];
1494                 } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) {
1495                         $nick = $matches[1];
1496                 } else {
1497                         return [];
1498                 }
1499
1500                 $data = [];
1501                 $data['url'] = 'https://twitter.com/' . $nick;
1502                 $data['addr'] = $nick . '@twitter.com';
1503                 $data['nick'] = $data['name'] = $nick;
1504                 $data['network'] = Protocol::TWITTER;
1505                 $data['baseurl'] = 'https://twitter.com';
1506
1507                 $curlResult = Network::curl($data['url'], false);
1508                 if (!$curlResult->isSuccess()) {
1509                         return [];
1510                 }
1511
1512                 $body = $curlResult->getBody();
1513                 $doc = new DOMDocument();
1514                 @$doc->loadHTML($body);
1515                 $xpath = new DOMXPath($doc);
1516
1517                 $list = $xpath->query('//img[@class]');
1518                 foreach ($list as $node) {
1519                         $img_attr = [];
1520                         if ($node->attributes->length) {
1521                                 foreach ($node->attributes as $attribute) {
1522                                         $img_attr[$attribute->name] = $attribute->value;
1523                                 }
1524                         }
1525
1526                         if (empty($img_attr['class'])) {
1527                                 continue;
1528                         }
1529
1530                         if (strpos($img_attr['class'], 'ProfileAvatar-image') !== false) {
1531                                 if (!empty($img_attr['src'])) {
1532                                         $data['photo'] = $img_attr['src'];
1533                                 }
1534                                 if (!empty($img_attr['alt'])) {
1535                                         $data['name'] = $img_attr['alt'];
1536                                 }
1537                         }
1538                 }
1539
1540                 return $data;
1541         }
1542
1543         /**
1544          * @brief Check page for feed link
1545          *
1546          * @param string $url Page link
1547          *
1548          * @return string feed link
1549          */
1550         private static function getFeedLink($url)
1551         {
1552                 $doc = new DOMDocument();
1553
1554                 if (!@$doc->loadHTMLFile($url)) {
1555                         return false;
1556                 }
1557
1558                 $xpath = new DomXPath($doc);
1559
1560                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1561                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1562                 if (!is_object($feeds)) {
1563                         return false;
1564                 }
1565
1566                 if ($feeds->length == 0) {
1567                         return false;
1568                 }
1569
1570                 $feed_url = "";
1571
1572                 foreach ($feeds as $feed) {
1573                         $attr = [];
1574                         foreach ($feed->attributes as $attribute) {
1575                                 $attr[$attribute->name] = trim($attribute->value);
1576                         }
1577
1578                         if ($feed_url == "") {
1579                                 $feed_url = $attr["href"];
1580                         }
1581                 }
1582
1583                 return $feed_url;
1584         }
1585
1586         /**
1587          * @brief Check for feed contact
1588          *
1589          * @param string  $url   Profile link
1590          * @param boolean $probe Do a probe if the page contains a feed link
1591          *
1592          * @return array feed data
1593          * @throws HTTPException\InternalServerErrorException
1594          */
1595         private static function feed($url, $probe = true)
1596         {
1597                 $curlResult = Network::curl($url);
1598                 if ($curlResult->isTimeout()) {
1599                         self::$istimeout = true;
1600                         return false;
1601                 }
1602                 $feed = $curlResult->getBody();
1603                 $dummy1 = $dummy2 = $dummy3 = null;
1604                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1605
1606                 if (!$feed_data) {
1607                         if (!$probe) {
1608                                 return false;
1609                         }
1610
1611                         $feed_url = self::getFeedLink($url);
1612
1613                         if (!$feed_url) {
1614                                 return false;
1615                         }
1616
1617                         return self::feed($feed_url, false);
1618                 }
1619
1620                 if (!empty($feed_data["header"]["author-name"])) {
1621                         $data["name"] = $feed_data["header"]["author-name"];
1622                 }
1623
1624                 if (!empty($feed_data["header"]["author-nick"])) {
1625                         $data["nick"] = $feed_data["header"]["author-nick"];
1626                 }
1627
1628                 if (!empty($feed_data["header"]["author-avatar"])) {
1629                         $data["photo"] = $feed_data["header"]["author-avatar"];
1630                 }
1631
1632                 if (!empty($feed_data["header"]["author-id"])) {
1633                         $data["alias"] = $feed_data["header"]["author-id"];
1634                 }
1635
1636                 $data["url"] = $url;
1637                 $data["poll"] = $url;
1638
1639                 if (!empty($feed_data["header"]["author-link"])) {
1640                         $data["baseurl"] = $feed_data["header"]["author-link"];
1641                 } else {
1642                         $data["baseurl"] = $data["url"];
1643                 }
1644
1645                 $data["network"] = Protocol::FEED;
1646
1647                 return $data;
1648         }
1649
1650         /**
1651          * @brief Check for mail contact
1652          *
1653          * @param string  $uri Profile link
1654          * @param integer $uid User ID
1655          *
1656          * @return array mail data
1657          * @throws \Exception
1658          */
1659         private static function mail($uri, $uid)
1660         {
1661                 if (!Network::isEmailDomainValid($uri)) {
1662                         return false;
1663                 }
1664
1665                 if ($uid == 0) {
1666                         return false;
1667                 }
1668
1669                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1670
1671                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1672                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1673                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1674
1675                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1676                         return false;
1677                 }
1678
1679                 $mailbox = Email::constructMailboxName($mailacct);
1680                 $password = '';
1681                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1682                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1683                 if (!$mbox) {
1684                         return false;
1685                 }
1686
1687                 $msgs = Email::poll($mbox, $uri);
1688                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1689
1690                 if (!count($msgs)) {
1691                         return false;
1692                 }
1693
1694                 $phost = substr($uri, strpos($uri, '@') + 1);
1695
1696                 $data = [];
1697                 $data["addr"]    = $uri;
1698                 $data["network"] = Protocol::MAIL;
1699                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1700                 $data["nick"]    = $data["name"];
1701                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1702                 $data["url"]     = 'mailto:'.$uri;
1703                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1704                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1705
1706                 $x = Email::messageMeta($mbox, $msgs[0]);
1707                 if (stristr($x[0]->from, $uri)) {
1708                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1709                 } elseif (stristr($x[0]->to, $uri)) {
1710                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1711                 }
1712                 if (isset($adr)) {
1713                         foreach ($adr as $feadr) {
1714                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1715                                         &&(strcasecmp($feadr->host, $phost) == 0)
1716                                         && (strlen($feadr->personal))
1717                                 ) {
1718                                         $personal = imap_mime_header_decode($feadr->personal);
1719                                         $data["name"] = "";
1720                                         foreach ($personal as $perspart) {
1721                                                 if ($perspart->charset != "default") {
1722                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1723                                                 } else {
1724                                                         $data["name"] .= $perspart->text;
1725                                                 }
1726                                         }
1727
1728                                         $data["name"] = Strings::escapeTags($data["name"]);
1729                                 }
1730                         }
1731                 }
1732                 if (!empty($mbox)) {
1733                         imap_close($mbox);
1734                 }
1735                 return $data;
1736         }
1737
1738         /**
1739          * @brief Mix two paths together to possibly fix missing parts
1740          *
1741          * @param string $avatar Path to the avatar
1742          * @param string $base   Another path that is hopefully complete
1743          *
1744          * @return string fixed avatar path
1745          * @throws \Exception
1746          */
1747         public static function fixAvatar($avatar, $base)
1748         {
1749                 $base_parts = parse_url($base);
1750
1751                 // Remove all parts that could create a problem
1752                 unset($base_parts['path']);
1753                 unset($base_parts['query']);
1754                 unset($base_parts['fragment']);
1755
1756                 $avatar_parts = parse_url($avatar);
1757
1758                 // Now we mix them
1759                 $parts = array_merge($base_parts, $avatar_parts);
1760
1761                 // And put them together again
1762                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1763                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1764                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1765                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1766                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1767                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1768
1769                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1770
1771                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1772
1773                 return $fixed;
1774         }
1775 }