src/Network/Probe.php

   1 <?php
   2 /**
   3  * @file src/Network/Probe.php
   4  */
   5 namespace Friendica\Network;
   6
   7 /**
   8  * @file src/Network/Probe.php
   9  * @brief Functions for probing URL
  10  */
  11
  12 use DOMDocument;
  13 use DomXPath;
  14 use Friendica\Core\Cache;
  15 use Friendica\Core\Config;
  16 use Friendica\Core\Logger;
  17 use Friendica\Core\Protocol;
  18 use Friendica\Core\System;
  19 use Friendica\Database\DBA;
  20 use Friendica\Model\Profile;
  21 use Friendica\Protocol\ActivityPub;
  22 use Friendica\Protocol\Email;
  23 use Friendica\Protocol\Feed;
  24 use Friendica\Util\Crypto;
  25 use Friendica\Util\DateTimeFormat;
  26 use Friendica\Util\Network;
  27 use Friendica\Util\Strings;
  28 use Friendica\Util\XML;
  29
  30 /**
  31  * @brief This class contain functions for probing URL
  32  *
  33  */
  34 class Probe
  35 {
  36         private static $baseurl;
  37         private static $istimeout;
  38
  39         /**
  40          * @brief Rearrange the array so that it always has the same order
  41          *
  42          * @param array $data Unordered data
  43          *
  44          * @return array Ordered data
  45          */
  46         private static function rearrangeData($data)
  47         {
  48                 $fields = ["name", "nick", "guid", "url", "addr", "alias",
  49                                 "photo", "community", "keywords", "location", "about",
  50                                 "batch", "notify", "poll", "request", "confirm", "poco",
  51                                 "priority", "network", "pubkey", "baseurl"];
  52
  53                 $newdata = [];
  54                 foreach ($fields as $field) {
  55                         if (isset($data[$field])) {
  56                                 $newdata[$field] = $data[$field];
  57                         } else {
  58                                 $newdata[$field] = "";
  59                         }
  60                 }
  61
  62                 // We don't use the "priority" field anymore and replace it with a dummy.
  63                 $newdata["priority"] = 0;
  64
  65                 return $newdata;
  66         }
  67
  68         /**
  69          * @brief Check if the hostname belongs to the own server
  70          *
  71          * @param string $host The hostname that is to be checked
  72          *
  73          * @return bool Does the testes hostname belongs to the own server?
  74          */
  75         private static function ownHost($host)
  76         {
  77                 $own_host = \get_app()->getHostName();
  78
  79                 $parts = parse_url($host);
  80
  81                 if (!isset($parts['scheme'])) {
  82                         $parts = parse_url('http://'.$host);
  83                 }
  84
  85                 if (!isset($parts['host'])) {
  86                         return false;
  87                 }
  88                 return $parts['host'] == $own_host;
  89         }
  90
  91         /**
  92          * @brief Probes for webfinger path via "host-meta"
  93          *
  94          * We have to check if the servers in the future still will offer this.
  95          * It seems as if it was dropped from the standard.
  96          *
  97          * @param string $host The host part of an url
  98          *
  99          * @return array with template and type of the webfinger template for JSON or XML
 100          * @throws HTTPException\InternalServerErrorException
 101          */
 102         private static function hostMeta($host)
 103         {
 104                 // Reset the static variable
 105                 self::$baseurl = '';
 106
 107                 $ssl_url = "https://".$host."/.well-known/host-meta";
 108                 $url = "http://".$host."/.well-known/host-meta";
 109
 110                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 111
 112                 Logger::log("Probing for ".$host, Logger::DEBUG);
 113                 $xrd = null;
 114
 115                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 116                 if ($curlResult->isSuccess()) {
 117                         $xml = $curlResult->getBody();
 118                         $xrd = XML::parseString($xml, false);
 119                         $host_url = 'https://'.$host;
 120                 }
 121
 122                 if (!is_object($xrd)) {
 123                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 124                         if ($curlResult->isTimeout()) {
 125                                 Logger::log("Probing timeout for " . $url, Logger::DEBUG);
 126                                 self::$istimeout = true;
 127                                 return false;
 128                         }
 129                         $xml = $curlResult->getBody();
 130                         $xrd = XML::parseString($xml, false);
 131                         $host_url = 'http://'.$host;
 132                 }
 133                 if (!is_object($xrd)) {
 134                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
 135                         return [];
 136                 }
 137
 138                 $links = XML::elementToArray($xrd);
 139                 if (!isset($links["xrd"]["link"])) {
 140                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
 141                         return [];
 142                 }
 143
 144                 $lrdd = [];
 145                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 146                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 147                 // compared to the XML webfinger. So this is commented out by now.
 148                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 149
 150                 foreach ($links["xrd"]["link"] as $value => $link) {
 151                         if (!empty($link["@attributes"])) {
 152                                 $attributes = $link["@attributes"];
 153                         } elseif ($value == "@attributes") {
 154                                 $attributes = $link;
 155                         } else {
 156                                 continue;
 157                         }
 158
 159                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 160                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 161
 162                                 $lrdd[$type] = $attributes["template"];
 163                         }
 164                 }
 165
 166                 self::$baseurl = $host_url;
 167
 168                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
 169
 170                 return $lrdd;
 171         }
 172
 173         /**
 174          * @brief Perform Webfinger lookup and return DFRN data
 175          *
 176          * Given an email style address, perform webfinger lookup and
 177          * return the resulting DFRN profile URL, or if no DFRN profile URL
 178          * is located, returns an OStatus subscription template (prefixed
 179          * with the string 'stat:' to identify it as on OStatus template).
 180          * If this isn't an email style address just return $webbie.
 181          * Return an empty string if email-style addresses but webfinger fails,
 182          * or if the resultant personal XRD doesn't contain a supported
 183          * subscription/friend-request attribute.
 184          *
 185          * amended 7/9/2011 to return an hcard which could save potentially loading
 186          * a lengthy content page to scrape dfrn attributes
 187          *
 188          * @param string $webbie    Address that should be probed
 189          * @param string $hcard_url Link to the hcard - is returned by reference
 190          *
 191          * @return string profile link
 192          * @throws HTTPException\InternalServerErrorException
 193          */
 194         public static function webfingerDfrn($webbie, &$hcard_url)
 195         {
 196                 $profile_link = '';
 197
 198                 $links = self::lrdd($webbie);
 199                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
 200                 if (!empty($links) && is_array($links)) {
 201                         foreach ($links as $link) {
 202                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 203                                         $profile_link = $link['@attributes']['href'];
 204                                 }
 205                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 206                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 207                                 }
 208                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 209                                         $hcard_url = $link['@attributes']['href'];
 210                                 }
 211                         }
 212                 }
 213                 return $profile_link;
 214         }
 215
 216         /**
 217          * @brief Check an URI for LRDD data
 218          *
 219          * this is a replacement for the "lrdd" function.
 220          * It isn't used in this class and has some redundancies in the code.
 221          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 222          *
 223          * @param string $uri Address that should be probed
 224          *
 225          * @return array uri data
 226          * @throws HTTPException\InternalServerErrorException
 227          */
 228         public static function lrdd($uri)
 229         {
 230                 $lrdd = self::hostMeta($uri);
 231                 $webfinger = null;
 232
 233                 if (is_bool($lrdd)) {
 234                         return [];
 235                 }
 236
 237                 if (!$lrdd) {
 238                         $parts = @parse_url($uri);
 239                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 240                                 return [];
 241                         }
 242
 243                         $host = $parts["host"];
 244                         if (!empty($parts["port"])) {
 245                                 $host .= ':'.$parts["port"];
 246                         }
 247
 248                         $path_parts = explode("/", trim($parts["path"], "/"));
 249
 250                         $nick = array_pop($path_parts);
 251
 252                         do {
 253                                 $lrdd = self::hostMeta($host);
 254                                 $host .= "/".array_shift($path_parts);
 255                         } while (!$lrdd && (sizeof($path_parts) > 0));
 256                 }
 257
 258                 if (!$lrdd) {
 259                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
 260                         return [];
 261                 }
 262
 263                 foreach ($lrdd as $type => $template) {
 264                         if ($webfinger) {
 265                                 continue;
 266                         }
 267
 268                         $path = str_replace('{uri}', urlencode($uri), $template);
 269                         $webfinger = self::webfinger($path, $type);
 270
 271                         if (!$webfinger && (strstr($uri, "@"))) {
 272                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 273                                 $webfinger = self::webfinger($path, $type);
 274                         }
 275
 276                         // Special treatment for Mastodon
 277                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 278                         // But the webfinger for this format fails.
 279                         if (!$webfinger && !empty($nick)) {
 280                                 // Mastodon uses a "@" as prefix for usernames in their url format
 281                                 $nick = ltrim($nick, '@');
 282
 283                                 $addr = $nick."@".$host;
 284
 285                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 286                                 $webfinger = self::webfinger($path, $type);
 287                         }
 288                 }
 289
 290                 if (!is_array($webfinger["links"])) {
 291                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
 292                         return false;
 293                 }
 294
 295                 $data = [];
 296
 297                 foreach ($webfinger["links"] as $link) {
 298                         $data[] = ["@attributes" => $link];
 299                 }
 300
 301                 if (is_array($webfinger["aliases"])) {
 302                         foreach ($webfinger["aliases"] as $alias) {
 303                                 $data[] = ["@attributes" =>
 304                                                         ["rel" => "alias",
 305                                                                 "href" => $alias]];
 306                         }
 307                 }
 308
 309                 return $data;
 310         }
 311
 312         /**
 313          * @brief Fetch information (protocol endpoints and user information) about a given uri
 314          *
 315          * @param string  $uri     Address that should be probed
 316          * @param string  $network Test for this specific network
 317          * @param integer $uid     User ID for the probe (only used for mails)
 318          * @param boolean $cache   Use cached values?
 319          *
 320          * @return array uri data
 321          * @throws HTTPException\InternalServerErrorException
 322          * @throws \ImagickException
 323          */
 324         public static function uri($uri, $network = '', $uid = -1, $cache = true)
 325         {
 326                 if ($cache) {
 327                         $result = Cache::get('Probe::uri:' . $network . ':' . $uri);
 328                         if (!is_null($result)) {
 329                                 return $result;
 330                         }
 331                 }
 332
 333                 if ($uid == -1) {
 334                         $uid = local_user();
 335                 }
 336
 337                 self::$istimeout = false;
 338
 339                 if ($network != Protocol::ACTIVITYPUB) {
 340                         $data = self::detect($uri, $network, $uid);
 341                 } else {
 342                         $data = null;
 343                 }
 344
 345                 // When the previous detection process had got a time out
 346                 // we could falsely detect a Friendica profile as AP profile.
 347                 if (!self::$istimeout) {
 348                         $ap_profile = ActivityPub::probeProfile($uri);
 349
 350                         if (!empty($ap_profile) && empty($network) && (defaults($data, 'network', '') != Protocol::DFRN)) {
 351                                 $data = $ap_profile;
 352                         }
 353                 } else {
 354                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
 355                 }
 356
 357                 if (!isset($data['url'])) {
 358                         $data['url'] = $uri;
 359                 }
 360
 361                 if (!empty($data['photo'])) {
 362                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink(defaults($data, 'baseurl', '')), Strings::normaliseLink($data['photo']));
 363                 } else {
 364                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 365                 }
 366
 367                 if (empty($data['name'])) {
 368                         if (!empty($data['nick'])) {
 369                                 $data['name'] = $data['nick'];
 370                         }
 371
 372                         if (empty($data['name'])) {
 373                                 $data['name'] = $data['url'];
 374                         }
 375                 }
 376
 377                 if (empty($data['nick'])) {
 378                         $data['nick'] = strtolower($data['name']);
 379
 380                         if (strpos($data['nick'], ' ')) {
 381                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 382                         }
 383                 }
 384
 385                 if (!empty(self::$baseurl)) {
 386                         $data['baseurl'] = self::$baseurl;
 387                 }
 388
 389                 if (empty($data['network'])) {
 390                         $data['network'] = Protocol::PHANTOM;
 391                 }
 392
 393                 $data = self::rearrangeData($data);
 394
 395                 // Only store into the cache if the value seems to be valid
 396                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 397                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 398                 }
 399
 400                 return $data;
 401         }
 402
 403         /**
 404          * @brief Checks if a profile url should be OStatus but only provides partial information
 405          *
 406          * @param array  $webfinger Webfinger data
 407          * @param string $lrdd      Path template for webfinger request
 408          * @param string $type      type
 409          *
 410          * @return array fixed webfinger data
 411          * @throws HTTPException\InternalServerErrorException
 412          */
 413         private static function fixOStatus($webfinger, $lrdd, $type)
 414         {
 415                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 416                         return $webfinger;
 417                 }
 418
 419                 $is_ostatus = false;
 420                 $has_key = false;
 421
 422                 foreach ($webfinger['links'] as $link) {
 423                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 424                                 $is_ostatus = true;
 425                         }
 426                         if ($link['rel'] == 'magic-public-key') {
 427                                 $has_key = true;
 428                         }
 429                 }
 430
 431                 if (!$is_ostatus || $has_key) {
 432                         return $webfinger;
 433                 }
 434
 435                 $url = Network::switchScheme($webfinger['subject']);
 436                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 437                 $webfinger2 = self::webfinger($path, $type);
 438
 439                 // Is the new webfinger detectable as OStatus?
 440                 if (self::ostatus($webfinger2, true)) {
 441                         $webfinger = $webfinger2;
 442                 }
 443
 444                 return $webfinger;
 445         }
 446
 447         /**
 448          * @brief Fetch information (protocol endpoints and user information) about a given uri
 449          *
 450          * This function is only called by the "uri" function that adds caching and rearranging of data.
 451          *
 452          * @param string  $uri     Address that should be probed
 453          * @param string  $network Test for this specific network
 454          * @param integer $uid     User ID for the probe (only used for mails)
 455          *
 456          * @return array uri data
 457          * @throws HTTPException\InternalServerErrorException
 458          */
 459         private static function detect($uri, $network, $uid)
 460         {
 461                 $parts = parse_url($uri);
 462
 463                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 464                         $host = $parts["host"];
 465                         if (!empty($parts["port"])) {
 466                                 $host .= ':'.$parts["port"];
 467                         }
 468
 469                         if ($host == 'twitter.com') {
 470                                 return ["network" => Protocol::TWITTER];
 471                         }
 472                         $lrdd = self::hostMeta($host);
 473
 474                         if (is_bool($lrdd)) {
 475                                 return [];
 476                         }
 477
 478                         $path_parts = explode("/", trim(defaults($parts, 'path', ''), "/"));
 479
 480                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 481                                 $host .= "/".array_shift($path_parts);
 482                                 $lrdd = self::hostMeta($host);
 483                         }
 484                         if (!$lrdd) {
 485                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 486                                 return self::feed($uri);
 487                         }
 488                         $nick = array_pop($path_parts);
 489
 490                         // Mastodon uses a "@" as prefix for usernames in their url format
 491                         $nick = ltrim($nick, '@');
 492
 493                         $addr = $nick."@".$host;
 494                 } elseif (strstr($uri, '@')) {
 495                         // If the URI starts with "mailto:" then jump directly to the mail detection
 496                         if (strpos($uri, 'mailto:') !== false) {
 497                                 $uri = str_replace('mailto:', '', $uri);
 498                                 return self::mail($uri, $uid);
 499                         }
 500
 501                         if ($network == Protocol::MAIL) {
 502                                 return self::mail($uri, $uid);
 503                         }
 504                         // Remove "acct:" from the URI
 505                         $uri = str_replace('acct:', '', $uri);
 506
 507                         $host = substr($uri, strpos($uri, '@') + 1);
 508                         $nick = substr($uri, 0, strpos($uri, '@'));
 509
 510                         if (strpos($uri, '@twitter.com')) {
 511                                 return ["network" => Protocol::TWITTER];
 512                         }
 513                         $lrdd = self::hostMeta($host);
 514
 515                         if (is_bool($lrdd)) {
 516                                 return [];
 517                         }
 518
 519                         if (!$lrdd) {
 520                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 521                                 return self::mail($uri, $uid);
 522                         }
 523                         $addr = $uri;
 524                 } else {
 525                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 526                         return false;
 527                 }
 528
 529                 $webfinger = false;
 530
 531                 /// @todo Do we need the prefix "acct:" or "acct://"?
 532
 533                 foreach ($lrdd as $type => $template) {
 534                         if ($webfinger) {
 535                                 continue;
 536                         }
 537
 538                         // At first try it with the given uri
 539                         $path = str_replace('{uri}', urlencode($uri), $template);
 540                         $webfinger = self::webfinger($path, $type);
 541
 542                         // Fix possible problems with GNU Social probing to wrong scheme
 543                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 544
 545                         // We cannot be sure that the detected address was correct, so we don't use the values
 546                         if ($webfinger && ($uri != $addr)) {
 547                                 $nick = "";
 548                                 $addr = "";
 549                         }
 550
 551                         // Try webfinger with the address (user@domain.tld)
 552                         if (!$webfinger) {
 553                                 $path = str_replace('{uri}', urlencode($addr), $template);
 554                                 $webfinger = self::webfinger($path, $type);
 555                         }
 556
 557                         // Mastodon needs to have it with "acct:"
 558                         if (!$webfinger) {
 559                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 560                                 $webfinger = self::webfinger($path, $type);
 561                         }
 562                 }
 563
 564                 if (!$webfinger) {
 565                         return self::feed($uri);
 566                 }
 567
 568                 $result = false;
 569
 570                 Logger::log("Probing ".$uri, Logger::DEBUG);
 571
 572                 if (in_array($network, ["", Protocol::DFRN])) {
 573                         $result = self::dfrn($webfinger);
 574                 }
 575                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 576                         $result = self::diaspora($webfinger);
 577                 }
 578                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 579                         $result = self::ostatus($webfinger);
 580                 }
 581                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 582                         $result = self::pumpio($webfinger, $addr);
 583                 }
 584                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 585                         $result = self::feed($uri);
 586                 } else {
 587                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 588                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 589                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 590                                 $result["nick"] = $nick;
 591                         }
 592
 593                         if (empty($result["addr"]) && ($addr != "")) {
 594                                 $result["addr"] = $addr;
 595                         }
 596                 }
 597
 598                 if (empty($result["network"])) {
 599                         $result["network"] = Protocol::PHANTOM;
 600                 }
 601
 602                 if (empty($result["url"])) {
 603                         $result["url"] = $uri;
 604                 }
 605
 606                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 607
 608                 if (empty($result["baseurl"])) {
 609                         $pos = strpos($result["url"], $host);
 610                         if ($pos) {
 611                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 612                         }
 613                 }
 614                 return $result;
 615         }
 616
 617         /**
 618          * @brief Perform a webfinger request.
 619          *
 620          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 621          *
 622          * @param string $url  Address that should be probed
 623          * @param string $type type
 624          *
 625          * @return array webfinger data
 626          * @throws HTTPException\InternalServerErrorException
 627          */
 628         private static function webfinger($url, $type)
 629         {
 630                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 631
 632                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 633                 if ($curlResult->isTimeout()) {
 634                         self::$istimeout = true;
 635                         return false;
 636                 }
 637                 $data = $curlResult->getBody();
 638
 639                 $webfinger = json_decode($data, true);
 640                 if (is_array($webfinger)) {
 641                         if (!isset($webfinger["links"])) {
 642                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 643                                 return false;
 644                         }
 645                         return $webfinger;
 646                 }
 647
 648                 // If it is not JSON, maybe it is XML
 649                 $xrd = XML::parseString($data, false);
 650                 if (!is_object($xrd)) {
 651                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 652                         return false;
 653                 }
 654
 655                 $xrd_arr = XML::elementToArray($xrd);
 656                 if (!isset($xrd_arr["xrd"]["link"])) {
 657                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 658                         return false;
 659                 }
 660
 661                 $webfinger = [];
 662
 663                 if (!empty($xrd_arr["xrd"]["subject"])) {
 664                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 665                 }
 666
 667                 if (!empty($xrd_arr["xrd"]["alias"])) {
 668                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 669                 }
 670
 671                 $webfinger["links"] = [];
 672
 673                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 674                         if (!empty($data["@attributes"])) {
 675                                 $attributes = $data["@attributes"];
 676                         } elseif ($value == "@attributes") {
 677                                 $attributes = $data;
 678                         } else {
 679                                 continue;
 680                         }
 681
 682                         $webfinger["links"][] = $attributes;
 683                 }
 684                 return $webfinger;
 685         }
 686
 687         /**
 688          * @brief Poll the Friendica specific noscrape page.
 689          *
 690          * "noscrape" is a faster alternative to fetch the data from the hcard.
 691          * This functionality was originally created for the directory.
 692          *
 693          * @param string $noscrape_url Link to the noscrape page
 694          * @param array  $data         The already fetched data
 695          *
 696          * @return array noscrape data
 697          * @throws HTTPException\InternalServerErrorException
 698          */
 699         private static function pollNoscrape($noscrape_url, $data)
 700         {
 701                 $curlResult = Network::curl($noscrape_url);
 702                 if ($curlResult->isTimeout()) {
 703                         self::$istimeout = true;
 704                         return false;
 705                 }
 706                 $content = $curlResult->getBody();
 707                 if (!$content) {
 708                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 709                         return false;
 710                 }
 711
 712                 $json = json_decode($content, true);
 713                 if (!is_array($json)) {
 714                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 715                         return false;
 716                 }
 717
 718                 if (!empty($json["fn"])) {
 719                         $data["name"] = $json["fn"];
 720                 }
 721
 722                 if (!empty($json["addr"])) {
 723                         $data["addr"] = $json["addr"];
 724                 }
 725
 726                 if (!empty($json["nick"])) {
 727                         $data["nick"] = $json["nick"];
 728                 }
 729
 730                 if (!empty($json["guid"])) {
 731                         $data["guid"] = $json["guid"];
 732                 }
 733
 734                 if (!empty($json["comm"])) {
 735                         $data["community"] = $json["comm"];
 736                 }
 737
 738                 if (!empty($json["tags"])) {
 739                         $keywords = implode(" ", $json["tags"]);
 740                         if ($keywords != "") {
 741                                 $data["keywords"] = $keywords;
 742                         }
 743                 }
 744
 745                 $location = Profile::formatLocation($json);
 746                 if ($location) {
 747                         $data["location"] = $location;
 748                 }
 749
 750                 if (!empty($json["about"])) {
 751                         $data["about"] = $json["about"];
 752                 }
 753
 754                 if (!empty($json["key"])) {
 755                         $data["pubkey"] = $json["key"];
 756                 }
 757
 758                 if (!empty($json["photo"])) {
 759                         $data["photo"] = $json["photo"];
 760                 }
 761
 762                 if (!empty($json["dfrn-request"])) {
 763                         $data["request"] = $json["dfrn-request"];
 764                 }
 765
 766                 if (!empty($json["dfrn-confirm"])) {
 767                         $data["confirm"] = $json["dfrn-confirm"];
 768                 }
 769
 770                 if (!empty($json["dfrn-notify"])) {
 771                         $data["notify"] = $json["dfrn-notify"];
 772                 }
 773
 774                 if (!empty($json["dfrn-poll"])) {
 775                         $data["poll"] = $json["dfrn-poll"];
 776                 }
 777
 778                 return $data;
 779         }
 780
 781         /**
 782          * @brief Check for valid DFRN data
 783          *
 784          * @param array $data DFRN data
 785          *
 786          * @return int Number of errors
 787          */
 788         public static function validDfrn($data)
 789         {
 790                 $errors = 0;
 791                 if (!isset($data['key'])) {
 792                         $errors ++;
 793                 }
 794                 if (!isset($data['dfrn-request'])) {
 795                         $errors ++;
 796                 }
 797                 if (!isset($data['dfrn-confirm'])) {
 798                         $errors ++;
 799                 }
 800                 if (!isset($data['dfrn-notify'])) {
 801                         $errors ++;
 802                 }
 803                 if (!isset($data['dfrn-poll'])) {
 804                         $errors ++;
 805                 }
 806                 return $errors;
 807         }
 808
 809         /**
 810          * @brief Fetch data from a DFRN profile page and via "noscrape"
 811          *
 812          * @param string $profile_link Link to the profile page
 813          *
 814          * @return array profile data
 815          * @throws HTTPException\InternalServerErrorException
 816          * @throws \ImagickException
 817          */
 818         public static function profile($profile_link)
 819         {
 820                 $data = [];
 821
 822                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
 823
 824                 // Fetch data via noscrape - this is faster
 825                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 826                 $data = self::pollNoscrape($noscrape_url, $data);
 827
 828                 if (!isset($data["notify"])
 829                         || !isset($data["confirm"])
 830                         || !isset($data["request"])
 831                         || !isset($data["poll"])
 832                         || !isset($data["name"])
 833                         || !isset($data["photo"])
 834                 ) {
 835                         $data = self::pollHcard($profile_link, $data, true);
 836                 }
 837
 838                 $prof_data = [];
 839
 840                 if (empty($data["addr"]) || empty($data["nick"])) {
 841                         $probe_data = self::uri($profile_link);
 842                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 843                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 844                 }
 845
 846                 $prof_data["addr"]         = $data["addr"];
 847                 $prof_data["nick"]         = $data["nick"];
 848                 $prof_data["dfrn-request"] = defaults($data, 'request', null);
 849                 $prof_data["dfrn-confirm"] = defaults($data, 'confirm', null);
 850                 $prof_data["dfrn-notify"]  = defaults($data, 'notify' , null);
 851                 $prof_data["dfrn-poll"]    = defaults($data, 'poll'   , null);
 852                 $prof_data["photo"]        = defaults($data, 'photo'  , null);
 853                 $prof_data["fn"]           = defaults($data, 'name'   , null);
 854                 $prof_data["key"]          = defaults($data, 'pubkey' , null);
 855
 856                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
 857
 858                 return $prof_data;
 859         }
 860
 861         /**
 862          * @brief Check for DFRN contact
 863          *
 864          * @param array $webfinger Webfinger data
 865          *
 866          * @return array DFRN data
 867          * @throws HTTPException\InternalServerErrorException
 868          */
 869         private static function dfrn($webfinger)
 870         {
 871                 $hcard_url = "";
 872                 $data = [];
 873                 // The array is reversed to take into account the order of preference for same-rel links
 874                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
 875                 foreach (array_reverse($webfinger["links"]) as $link) {
 876                         if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) {
 877                                 $data["network"] = Protocol::DFRN;
 878                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
 879                                 $data["poll"] = $link["href"];
 880                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
 881                                 $data["url"] = $link["href"];
 882                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
 883                                 $hcard_url = $link["href"];
 884                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
 885                                 $data["poco"] = $link["href"];
 886                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
 887                                 $data["photo"] = $link["href"];
 888                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
 889                                 $data["baseurl"] = trim($link["href"], '/');
 890                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
 891                                 $data["guid"] = $link["href"];
 892                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
 893                                 $data["pubkey"] = base64_decode($link["href"]);
 894
 895                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 896                                 if (strstr($data["pubkey"], 'RSA ')) {
 897                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 898                                 }
 899                         }
 900                 }
 901
 902                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 903                         foreach ($webfinger["aliases"] as $alias) {
 904                                 if (empty($data["url"]) && !strstr($alias, "@")) {
 905                                         $data["url"] = $alias;
 906                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
 907                                         $data["alias"] = $alias;
 908                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 909                                         $data["addr"] = substr($alias, 5);
 910                                 }
 911                         }
 912                 }
 913
 914                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 915                         $data["addr"] = substr($webfinger["subject"], 5);
 916                 }
 917
 918                 if (!isset($data["network"]) || ($hcard_url == "")) {
 919                         return false;
 920                 }
 921
 922                 // Fetch data via noscrape - this is faster
 923                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
 924                 $data = self::pollNoscrape($noscrape_url, $data);
 925
 926                 if (isset($data["notify"])
 927                         && isset($data["confirm"])
 928                         && isset($data["request"])
 929                         && isset($data["poll"])
 930                         && isset($data["name"])
 931                         && isset($data["photo"])
 932                 ) {
 933                         return $data;
 934                 }
 935
 936                 $data = self::pollHcard($hcard_url, $data, true);
 937
 938                 return $data;
 939         }
 940
 941         /**
 942          * @brief Poll the hcard page (Diaspora and Friendica specific)
 943          *
 944          * @param string  $hcard_url Link to the hcard page
 945          * @param array   $data      The already fetched data
 946          * @param boolean $dfrn      Poll DFRN specific data
 947          *
 948          * @return array hcard data
 949          * @throws HTTPException\InternalServerErrorException
 950          */
 951         private static function pollHcard($hcard_url, $data, $dfrn = false)
 952         {
 953                 $curlResult = Network::curl($hcard_url);
 954                 if ($curlResult->isTimeout()) {
 955                         self::$istimeout = true;
 956                         return false;
 957                 }
 958                 $content = $curlResult->getBody();
 959                 if (!$content) {
 960                         return false;
 961                 }
 962
 963                 $doc = new DOMDocument();
 964                 if (!@$doc->loadHTML($content)) {
 965                         return false;
 966                 }
 967
 968                 $xpath = new DomXPath($doc);
 969
 970                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 971                 if (!is_object($vcards)) {
 972                         return false;
 973                 }
 974
 975                 if (!isset($data["baseurl"])) {
 976                         $data["baseurl"] = "";
 977                 }
 978
 979                 if ($vcards->length > 0) {
 980                         $vcard = $vcards->item(0);
 981
 982                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 983                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 984                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 985                         if (($search->length > 0) && empty($data["guid"])) {
 986                                 $data["guid"] = $search->item(0)->nodeValue;
 987                         }
 988
 989                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 990                         if ($search->length > 0) {
 991                                 $data["nick"] = $search->item(0)->nodeValue;
 992                         }
 993
 994                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 995                         if ($search->length > 0) {
 996                                 $data["name"] = $search->item(0)->nodeValue;
 997                         }
 998
 999                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1000                         if ($search->length > 0) {
1001                                 $data["searchable"] = $search->item(0)->nodeValue;
1002                         }
1003
1004                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1005                         if ($search->length > 0) {
1006                                 $data["pubkey"] = $search->item(0)->nodeValue;
1007                                 if (strstr($data["pubkey"], 'RSA ')) {
1008                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1009                                 }
1010                         }
1011
1012                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1013                         if ($search->length > 0) {
1014                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1015                         }
1016                 }
1017
1018                 $avatar = [];
1019                 if (!empty($vcard)) {
1020                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1021                         foreach ($photos as $photo) {
1022                                 $attr = [];
1023                                 foreach ($photo->attributes as $attribute) {
1024                                         $attr[$attribute->name] = trim($attribute->value);
1025                                 }
1026
1027                                 if (isset($attr["src"]) && isset($attr["width"])) {
1028                                         $avatar[$attr["width"]] = $attr["src"];
1029                                 }
1030
1031                                 // We don't have a width. So we just take everything that we got.
1032                                 // This is a Hubzilla workaround which doesn't send a width.
1033                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1034                                         $avatar[] = $attr["src"];
1035                                 }
1036                         }
1037                 }
1038
1039                 if (sizeof($avatar)) {
1040                         ksort($avatar);
1041                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1042                 }
1043
1044                 if ($dfrn) {
1045                         // Poll DFRN specific data
1046                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1047                         if ($search->length > 0) {
1048                                 foreach ($search as $link) {
1049                                         //$data["request"] = $search->item(0)->nodeValue;
1050                                         $attr = [];
1051                                         foreach ($link->attributes as $attribute) {
1052                                                 $attr[$attribute->name] = trim($attribute->value);
1053                                         }
1054
1055                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1056                                 }
1057                         }
1058
1059                         // Older Friendica versions had used the "uid" field differently than newer versions
1060                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1061                                 unset($data["guid"]);
1062                         }
1063                 }
1064
1065
1066                 return $data;
1067         }
1068
1069         /**
1070          * @brief Check for Diaspora contact
1071          *
1072          * @param array $webfinger Webfinger data
1073          *
1074          * @return array Diaspora data
1075          * @throws HTTPException\InternalServerErrorException
1076          */
1077         private static function diaspora($webfinger)
1078         {
1079                 $hcard_url = "";
1080                 $data = [];
1081                 // The array is reversed to take into account the order of preference for same-rel links
1082                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1083                 foreach (array_reverse($webfinger["links"]) as $link) {
1084                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1085                                 $hcard_url = $link["href"];
1086                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1087                                 $data["baseurl"] = trim($link["href"], '/');
1088                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1089                                 $data["guid"] = $link["href"];
1090                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1091                                 $data["url"] = $link["href"];
1092                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1093                                 $data["poll"] = $link["href"];
1094                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1095                                 $data["poco"] = $link["href"];
1096                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1097                                 $data["notify"] = $link["href"];
1098                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1099                                 $data["pubkey"] = base64_decode($link["href"]);
1100
1101                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1102                                 if (strstr($data["pubkey"], 'RSA ')) {
1103                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1104                                 }
1105                         }
1106                 }
1107
1108                 if (!isset($data["url"]) || ($hcard_url == "")) {
1109                         return false;
1110                 }
1111
1112                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1113                         foreach ($webfinger["aliases"] as $alias) {
1114                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1115                                         $data["alias"] = $alias;
1116                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1117                                         $data["addr"] = substr($alias, 5);
1118                                 }
1119                         }
1120                 }
1121
1122                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1123                         $data["addr"] = substr($webfinger["subject"], 5);
1124                 }
1125
1126                 // Fetch further information from the hcard
1127                 $data = self::pollHcard($hcard_url, $data);
1128
1129                 if (!$data) {
1130                         return false;
1131                 }
1132
1133                 if (isset($data["url"])
1134                         && isset($data["guid"])
1135                         && isset($data["baseurl"])
1136                         && isset($data["pubkey"])
1137                         && ($hcard_url != "")
1138                 ) {
1139                         $data["network"] = Protocol::DIASPORA;
1140
1141                         // The Diaspora handle must always be lowercase
1142                         if (!empty($data["addr"])) {
1143                                 $data["addr"] = strtolower($data["addr"]);
1144                         }
1145
1146                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1147                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1148                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1149                 } else {
1150                         return false;
1151                 }
1152
1153                 return $data;
1154         }
1155
1156         /**
1157          * @brief Check for OStatus contact
1158          *
1159          * @param array $webfinger Webfinger data
1160          * @param bool  $short     Short detection mode
1161          *
1162          * @return array|bool OStatus data or "false" on error or "true" on short mode
1163          * @throws HTTPException\InternalServerErrorException
1164          */
1165         private static function ostatus($webfinger, $short = false)
1166         {
1167                 $data = [];
1168
1169                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1170                         foreach ($webfinger["aliases"] as $alias) {
1171                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1172                                         $data["addr"] = str_replace('acct:', '', $alias);
1173                                 }
1174                         }
1175                 }
1176
1177                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1178                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1179                 ) {
1180                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1181                 }
1182
1183                 if (is_array($webfinger["links"])) {
1184                         // The array is reversed to take into account the order of preference for same-rel links
1185                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1186                         foreach (array_reverse($webfinger["links"]) as $link) {
1187                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1188                                         && (defaults($link, "type", "") == "text/html")
1189                                         && ($link["href"] != "")
1190                                 ) {
1191                                         $data["url"] = $link["href"];
1192                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1193                                         $data["notify"] = $link["href"];
1194                                 } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1195                                         $data["poll"] = $link["href"];
1196                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1197                                         $pubkey = $link["href"];
1198
1199                                         if (substr($pubkey, 0, 5) === 'data:') {
1200                                                 if (strstr($pubkey, ',')) {
1201                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1202                                                 } else {
1203                                                         $pubkey = substr($pubkey, 5);
1204                                                 }
1205                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1206                                                 $curlResult = Network::curl($pubkey);
1207                                                 if ($curlResult->isTimeout()) {
1208                                                         self::$istimeout = true;
1209                                                         return false;
1210                                                 }
1211                                                 $pubkey = $curlResult->getBody();
1212                                         }
1213
1214                                         $key = explode(".", $pubkey);
1215
1216                                         if (sizeof($key) >= 3) {
1217                                                 $m = Strings::base64UrlDecode($key[1]);
1218                                                 $e = Strings::base64UrlDecode($key[2]);
1219                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1220                                         }
1221                                 }
1222                         }
1223                 }
1224
1225                 if (isset($data["notify"]) && isset($data["pubkey"])
1226                         && isset($data["poll"])
1227                         && isset($data["url"])
1228                 ) {
1229                         $data["network"] = Protocol::OSTATUS;
1230                 } else {
1231                         return false;
1232                 }
1233
1234                 if ($short) {
1235                         return true;
1236                 }
1237
1238                 // Fetch all additional data from the feed
1239                 $curlResult = Network::curl($data["poll"]);
1240                 if ($curlResult->isTimeout()) {
1241                         self::$istimeout = true;
1242                         return false;
1243                 }
1244                 $feed = $curlResult->getBody();
1245                 $dummy1 = null;
1246                 $dummy2 = null;
1247                 $dummy2 = null;
1248                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1249                 if (!$feed_data) {
1250                         return false;
1251                 }
1252
1253                 if (!empty($feed_data["header"]["author-name"])) {
1254                         $data["name"] = $feed_data["header"]["author-name"];
1255                 }
1256                 if (!empty($feed_data["header"]["author-nick"])) {
1257                         $data["nick"] = $feed_data["header"]["author-nick"];
1258                 }
1259                 if (!empty($feed_data["header"]["author-avatar"])) {
1260                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1261                 }
1262                 if (!empty($feed_data["header"]["author-id"])) {
1263                         $data["alias"] = $feed_data["header"]["author-id"];
1264                 }
1265                 if (!empty($feed_data["header"]["author-location"])) {
1266                         $data["location"] = $feed_data["header"]["author-location"];
1267                 }
1268                 if (!empty($feed_data["header"]["author-about"])) {
1269                         $data["about"] = $feed_data["header"]["author-about"];
1270                 }
1271                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1272                 // So we take the value that we just fetched, although the other one worked as well
1273                 if (!empty($feed_data["header"]["author-link"])) {
1274                         $data["url"] = $feed_data["header"]["author-link"];
1275                 }
1276
1277                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1278                         $data['url'] = $data["alias"];
1279                         $data["alias"] = '';
1280                 }
1281
1282                 /// @todo Fetch location and "about" from the feed as well
1283                 return $data;
1284         }
1285
1286         /**
1287          * @brief Fetch data from a pump.io profile page
1288          *
1289          * @param string $profile_link Link to the profile page
1290          *
1291          * @return array profile data
1292          */
1293         private static function pumpioProfileData($profile_link)
1294         {
1295                 $doc = new DOMDocument();
1296                 if (!@$doc->loadHTMLFile($profile_link)) {
1297                         return false;
1298                 }
1299
1300                 $xpath = new DomXPath($doc);
1301
1302                 $data = [];
1303
1304                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1305
1306                 if ($data["name"] == '') {
1307                         // This is ugly - but pump.io doesn't seem to know a better way for it
1308                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1309                         $pos = strpos($data["name"], chr(10));
1310                         if ($pos) {
1311                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1312                         }
1313                 }
1314
1315                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1316
1317                 if ($data["location"] == '') {
1318                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1319                 }
1320
1321                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1322
1323                 if ($data["about"] == '') {
1324                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1325                 }
1326
1327                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1328                 if (!$avatar) {
1329                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1330                 }
1331                 if ($avatar) {
1332                         foreach ($avatar->attributes as $attribute) {
1333                                 if ($attribute->name == "src") {
1334                                         $data["photo"] = trim($attribute->value);
1335                                 }
1336                         }
1337                 }
1338
1339                 return $data;
1340         }
1341
1342         /**
1343          * @brief Check for pump.io contact
1344          *
1345          * @param array $webfinger Webfinger data
1346          *
1347          * @param       $addr
1348          * @return array pump.io data
1349          */
1350         private static function pumpio($webfinger, $addr)
1351         {
1352                 $data = [];
1353                 // The array is reversed to take into account the order of preference for same-rel links
1354                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1355                 foreach (array_reverse($webfinger["links"]) as $link) {
1356                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1357                                 && (defaults($link, "type", "") == "text/html")
1358                                 && ($link["href"] != "")
1359                         ) {
1360                                 $data["url"] = $link["href"];
1361                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1362                                 $data["notify"] = $link["href"];
1363                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1364                                 $data["poll"] = $link["href"];
1365                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1366                                 $data["dialback"] = $link["href"];
1367                         }
1368                 }
1369                 if (isset($data["poll"]) && isset($data["notify"])
1370                         && isset($data["dialback"])
1371                         && isset($data["url"])
1372                 ) {
1373                         // by now we use these fields only for the network type detection
1374                         // So we unset all data that isn't used at the moment
1375                         unset($data["dialback"]);
1376
1377                         $data["network"] = Protocol::PUMPIO;
1378                 } else {
1379                         return false;
1380                 }
1381
1382                 $profile_data = self::pumpioProfileData($data["url"]);
1383
1384                 if (!$profile_data) {
1385                         return false;
1386                 }
1387
1388                 $data = array_merge($data, $profile_data);
1389
1390                 if (($addr != '') && ($data['name'] != '')) {
1391                         $name = trim(str_replace($addr, '', $data['name']));
1392                         if ($name != '') {
1393                                 $data['name'] = $name;
1394                         }
1395                 }
1396
1397                 return $data;
1398         }
1399
1400         /**
1401          * @brief Check page for feed link
1402          *
1403          * @param string $url Page link
1404          *
1405          * @return string feed link
1406          */
1407         private static function getFeedLink($url)
1408         {
1409                 $doc = new DOMDocument();
1410
1411                 if (!@$doc->loadHTMLFile($url)) {
1412                         return false;
1413                 }
1414
1415                 $xpath = new DomXPath($doc);
1416
1417                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1418                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1419                 if (!is_object($feeds)) {
1420                         return false;
1421                 }
1422
1423                 if ($feeds->length == 0) {
1424                         return false;
1425                 }
1426
1427                 $feed_url = "";
1428
1429                 foreach ($feeds as $feed) {
1430                         $attr = [];
1431                         foreach ($feed->attributes as $attribute) {
1432                                 $attr[$attribute->name] = trim($attribute->value);
1433                         }
1434
1435                         if ($feed_url == "") {
1436                                 $feed_url = $attr["href"];
1437                         }
1438                 }
1439
1440                 return $feed_url;
1441         }
1442
1443         /**
1444          * @brief Check for feed contact
1445          *
1446          * @param string  $url   Profile link
1447          * @param boolean $probe Do a probe if the page contains a feed link
1448          *
1449          * @return array feed data
1450          * @throws HTTPException\InternalServerErrorException
1451          */
1452         private static function feed($url, $probe = true)
1453         {
1454                 $curlResult = Network::curl($url);
1455                 if ($curlResult->isTimeout()) {
1456                         self::$istimeout = true;
1457                         return false;
1458                 }
1459                 $feed = $curlResult->getBody();
1460                 $dummy1 = $dummy2 = $dummy3 = null;
1461                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1462
1463                 if (!$feed_data) {
1464                         if (!$probe) {
1465                                 return false;
1466                         }
1467
1468                         $feed_url = self::getFeedLink($url);
1469
1470                         if (!$feed_url) {
1471                                 return false;
1472                         }
1473
1474                         return self::feed($feed_url, false);
1475                 }
1476
1477                 if (!empty($feed_data["header"]["author-name"])) {
1478                         $data["name"] = $feed_data["header"]["author-name"];
1479                 }
1480
1481                 if (!empty($feed_data["header"]["author-nick"])) {
1482                         $data["nick"] = $feed_data["header"]["author-nick"];
1483                 }
1484
1485                 if (!empty($feed_data["header"]["author-avatar"])) {
1486                         $data["photo"] = $feed_data["header"]["author-avatar"];
1487                 }
1488
1489                 if (!empty($feed_data["header"]["author-id"])) {
1490                         $data["alias"] = $feed_data["header"]["author-id"];
1491                 }
1492
1493                 $data["url"] = $url;
1494                 $data["poll"] = $url;
1495
1496                 if (!empty($feed_data["header"]["author-link"])) {
1497                         $data["baseurl"] = $feed_data["header"]["author-link"];
1498                 } else {
1499                         $data["baseurl"] = $data["url"];
1500                 }
1501
1502                 $data["network"] = Protocol::FEED;
1503
1504                 return $data;
1505         }
1506
1507         /**
1508          * @brief Check for mail contact
1509          *
1510          * @param string  $uri Profile link
1511          * @param integer $uid User ID
1512          *
1513          * @return array mail data
1514          * @throws \Exception
1515          */
1516         private static function mail($uri, $uid)
1517         {
1518                 if (!Network::isEmailDomainValid($uri)) {
1519                         return false;
1520                 }
1521
1522                 if ($uid == 0) {
1523                         return false;
1524                 }
1525
1526                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1527
1528                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1529                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1530                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1531
1532                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1533                         return false;
1534                 }
1535
1536                 $mailbox = Email::constructMailboxName($mailacct);
1537                 $password = '';
1538                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1539                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1540                 if (!$mbox) {
1541                         return false;
1542                 }
1543
1544                 $msgs = Email::poll($mbox, $uri);
1545                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1546
1547                 if (!count($msgs)) {
1548                         return false;
1549                 }
1550
1551                 $phost = substr($uri, strpos($uri, '@') + 1);
1552
1553                 $data = [];
1554                 $data["addr"]    = $uri;
1555                 $data["network"] = Protocol::MAIL;
1556                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1557                 $data["nick"]    = $data["name"];
1558                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1559                 $data["url"]     = 'mailto:'.$uri;
1560                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1561                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1562
1563                 $x = Email::messageMeta($mbox, $msgs[0]);
1564                 if (stristr($x[0]->from, $uri)) {
1565                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1566                 } elseif (stristr($x[0]->to, $uri)) {
1567                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1568                 }
1569                 if (isset($adr)) {
1570                         foreach ($adr as $feadr) {
1571                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1572                                         &&(strcasecmp($feadr->host, $phost) == 0)
1573                                         && (strlen($feadr->personal))
1574                                 ) {
1575                                         $personal = imap_mime_header_decode($feadr->personal);
1576                                         $data["name"] = "";
1577                                         foreach ($personal as $perspart) {
1578                                                 if ($perspart->charset != "default") {
1579                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1580                                                 } else {
1581                                                         $data["name"] .= $perspart->text;
1582                                                 }
1583                                         }
1584
1585                                         $data["name"] = Strings::escapeTags($data["name"]);
1586                                 }
1587                         }
1588                 }
1589                 if (!empty($mbox)) {
1590                         imap_close($mbox);
1591                 }
1592                 return $data;
1593         }
1594
1595         /**
1596          * @brief Mix two paths together to possibly fix missing parts
1597          *
1598          * @param string $avatar Path to the avatar
1599          * @param string $base   Another path that is hopefully complete
1600          *
1601          * @return string fixed avatar path
1602          * @throws \Exception
1603          */
1604         public static function fixAvatar($avatar, $base)
1605         {
1606                 $base_parts = parse_url($base);
1607
1608                 // Remove all parts that could create a problem
1609                 unset($base_parts['path']);
1610                 unset($base_parts['query']);
1611                 unset($base_parts['fragment']);
1612
1613                 $avatar_parts = parse_url($avatar);
1614
1615                 // Now we mix them
1616                 $parts = array_merge($base_parts, $avatar_parts);
1617
1618                 // And put them together again
1619                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1620                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1621                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1622                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1623                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1624                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1625
1626                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1627
1628                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1629
1630                 return $fixed;
1631         }
1632 }