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 Friendica\Core\Cache;
  14 use Friendica\Core\Config;
  15 use Friendica\Core\Logger;
  16 use Friendica\Core\Protocol;
  17 use Friendica\Core\System;
  18 use Friendica\Database\DBA;
  19 use Friendica\Model\Contact;
  20 use Friendica\Model\Profile;
  21 use Friendica\Protocol\Email;
  22 use Friendica\Protocol\Feed;
  23 use Friendica\Protocol\ActivityPub;
  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 use DomXPath;
  30
  31 /**
  32  * @brief This class contain functions for probing URL
  33  *
  34  */
  35 class Probe
  36 {
  37         private static $baseurl;
  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                 $redirects = 0;
 112
 113                 Logger::log("Probing for ".$host, Logger::DEBUG);
 114                 $xrd = null;
 115
 116                 $curlResult = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 117                 if ($curlResult->isSuccess()) {
 118                         $xml = $curlResult->getBody();
 119                         $xrd = XML::parseString($xml, false);
 120                         $host_url = 'https://'.$host;
 121                 }
 122
 123                 if (!is_object($xrd)) {
 124                         $curlResult = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 125                         if ($curlResult->isTimeout()) {
 126                                 Logger::log("Probing timeout for " . $url, Logger::DEBUG);
 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 = $link;
 155                         } else {
 156                                 continue;
 157                         }
 158
 159                         if (($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 (count($links)) {
 201                         foreach ($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                         $uid = local_user();
 335                 }
 336
 337                 if ($network != Protocol::ACTIVITYPUB) {
 338                         $data = self::detect($uri, $network, $uid);
 339                 } else {
 340                         $data = null;
 341                 }
 342
 343                 $ap_profile = ActivityPub::probeProfile($uri);
 344
 345                 if (!empty($ap_profile) && (defaults($data, 'network', '') != Protocol::DFRN)) {
 346                         $data = $ap_profile;
 347                 }
 348
 349                 if (!isset($data['url'])) {
 350                         $data['url'] = $uri;
 351                 }
 352
 353                 if (!empty($data['photo'])) {
 354                         $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink(defaults($data, 'baseurl', '')), Strings::normaliseLink($data['photo']));
 355                 } else {
 356                         $data['photo'] = System::baseUrl() . '/images/person-300.jpg';
 357                 }
 358
 359                 if (empty($data['name'])) {
 360                         if (!empty($data['nick'])) {
 361                                 $data['name'] = $data['nick'];
 362                         }
 363
 364                         if (empty($data['name'])) {
 365                                 $data['name'] = $data['url'];
 366                         }
 367                 }
 368
 369                 if (empty($data['nick'])) {
 370                         $data['nick'] = strtolower($data['name']);
 371
 372                         if (strpos($data['nick'], ' ')) {
 373                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 374                         }
 375                 }
 376
 377                 if (!empty(self::$baseurl)) {
 378                         $data['baseurl'] = self::$baseurl;
 379                 }
 380
 381                 if (empty($data['network'])) {
 382                         $data['network'] = Protocol::PHANTOM;
 383                 }
 384
 385                 $data = self::rearrangeData($data);
 386
 387                 // Only store into the cache if the value seems to be valid
 388                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 389                         Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY);
 390
 391                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 392                         /// The biggest problem is the avatar picture that could have a reduced image size.
 393                         /// It should only be updated if the existing picture isn't existing anymore.
 394                         /// We only update the contact when it is no probing for a specific network.
 395                         if (($data['network'] != Protocol::FEED)
 396                                 && ($network == '')
 397                                 && $data['name']
 398                                 && $data['nick']
 399                                 && $data['url']
 400                                 && $data['addr']
 401                                 && $data['poll']
 402                         ) {
 403                                 $fields = [
 404                                         'name' => $data['name'],
 405                                         'nick' => $data['nick'],
 406                                         'url' => $data['url'],
 407                                         'addr' => $data['addr'],
 408                                         'photo' => $data['photo'],
 409                                         'keywords' => $data['keywords'],
 410                                         'location' => $data['location'],
 411                                         'about' => $data['about'],
 412                                         'notify' => $data['notify'],
 413                                         'network' => $data['network'],
 414                                         'server_url' => $data['baseurl']
 415                                 ];
 416
 417                                 // This doesn't cover the case when a community isn't a community anymore
 418                                 if (!empty($data['community']) && $data['community']) {
 419                                         $fields['community'] = $data['community'];
 420                                         $fields['contact-type'] = Contact::ACCOUNT_TYPE_COMMUNITY;
 421                                 }
 422
 423                                 $fieldnames = [];
 424
 425                                 foreach ($fields as $key => $val) {
 426                                         if (empty($val)) {
 427                                                 unset($fields[$key]);
 428                                         } else {
 429                                                 $fieldnames[] = $key;
 430                                         }
 431                                 }
 432
 433                                 $fields['updated'] = DateTimeFormat::utcNow();
 434
 435                                 $condition = ['nurl' => Strings::normaliseLink($data['url'])];
 436
 437                                 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
 438
 439                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 440                                 // In difference to the public contacts we want to have every contact
 441                                 // in the world in our global contacts.
 442                                 if (!$old_fields) {
 443                                         $old_fields = true;
 444
 445                                         // These values have to be set only on insert
 446                                         $fields['photo'] = $data['photo'];
 447                                         $fields['created'] = DateTimeFormat::utcNow();
 448                                 }
 449
 450                                 DBA::update('gcontact', $fields, $condition, $old_fields);
 451
 452                                 $fields = [
 453                                         'name' => $data['name'],
 454                                         'nick' => $data['nick'],
 455                                         'url' => $data['url'],
 456                                         'addr' => $data['addr'],
 457                                         'alias' => $data['alias'],
 458                                         'keywords' => $data['keywords'],
 459                                         'location' => $data['location'],
 460                                         'about' => $data['about'],
 461                                         'batch' => $data['batch'],
 462                                         'notify' => $data['notify'],
 463                                         'poll' => $data['poll'],
 464                                         'request' => $data['request'],
 465                                         'confirm' => $data['confirm'],
 466                                         'poco' => $data['poco'],
 467                                         'network' => $data['network'],
 468                                         'pubkey' => $data['pubkey'],
 469                                         'priority' => $data['priority'],
 470                                         'writable' => true,
 471                                         'rel' => Contact::SHARING
 472                                 ];
 473
 474                                 $fieldnames = [];
 475
 476                                 foreach ($fields as $key => $val) {
 477                                         if (empty($val)) {
 478                                                 unset($fields[$key]);
 479                                         } else {
 480                                                 $fieldnames[] = $key;
 481                                         }
 482                                 }
 483
 484                                 $condition = ['nurl' => Strings::normaliseLink($data['url']), 'self' => false, 'uid' => 0];
 485
 486                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 487                                 // This won't trigger an insert. This is intended, since we only need
 488                                 // public contacts for everyone we store items from.
 489                                 // We don't need to store every contact on the planet.
 490                                 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 491
 492                                 $fields['name-date'] = DateTimeFormat::utcNow();
 493                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 494                                 $fields['success_update'] = DateTimeFormat::utcNow();
 495
 496                                 DBA::update('contact', $fields, $condition, $old_fields);
 497                         }
 498                 }
 499
 500                 return $data;
 501         }
 502
 503         /**
 504          * @brief Switch the scheme of an url between http and https
 505          *
 506          * @param string $url URL
 507          *
 508          * @return string switched URL
 509          */
 510         private static function switchScheme($url)
 511         {
 512                 $parts = parse_url($url);
 513
 514                 if (!isset($parts['scheme'])) {
 515                         return $url;
 516                 }
 517
 518                 if ($parts['scheme'] == 'http') {
 519                         $url = str_replace('http://', 'https://', $url);
 520                 } elseif ($parts['scheme'] == 'https') {
 521                         $url = str_replace('https://', 'http://', $url);
 522                 }
 523
 524                 return $url;
 525         }
 526
 527         /**
 528          * @brief Checks if a profile url should be OStatus but only provides partial information
 529          *
 530          * @param array  $webfinger Webfinger data
 531          * @param string $lrdd      Path template for webfinger request
 532          * @param string $type      type
 533          *
 534          * @return array fixed webfinger data
 535          * @throws HTTPException\InternalServerErrorException
 536          */
 537         private static function fixOStatus($webfinger, $lrdd, $type)
 538         {
 539                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 540                         return $webfinger;
 541                 }
 542
 543                 $is_ostatus = false;
 544                 $has_key = false;
 545
 546                 foreach ($webfinger['links'] as $link) {
 547                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 548                                 $is_ostatus = true;
 549                         }
 550                         if ($link['rel'] == 'magic-public-key') {
 551                                 $has_key = true;
 552                         }
 553                 }
 554
 555                 if (!$is_ostatus || $has_key) {
 556                         return $webfinger;
 557                 }
 558
 559                 $url = self::switchScheme($webfinger['subject']);
 560                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 561                 $webfinger2 = self::webfinger($path, $type);
 562
 563                 // Is the new webfinger detectable as OStatus?
 564                 if (self::ostatus($webfinger2, true)) {
 565                         $webfinger = $webfinger2;
 566                 }
 567
 568                 return $webfinger;
 569         }
 570
 571         /**
 572          * @brief Fetch information (protocol endpoints and user information) about a given uri
 573          *
 574          * This function is only called by the "uri" function that adds caching and rearranging of data.
 575          *
 576          * @param string  $uri     Address that should be probed
 577          * @param string  $network Test for this specific network
 578          * @param integer $uid     User ID for the probe (only used for mails)
 579          *
 580          * @return array uri data
 581          * @throws HTTPException\InternalServerErrorException
 582          */
 583         private static function detect($uri, $network, $uid)
 584         {
 585                 $parts = parse_url($uri);
 586
 587                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
 588                         $host = $parts["host"];
 589                         if (!empty($parts["port"])) {
 590                                 $host .= ':'.$parts["port"];
 591                         }
 592
 593                         if ($host == 'twitter.com') {
 594                                 return ["network" => Protocol::TWITTER];
 595                         }
 596                         $lrdd = self::hostMeta($host);
 597
 598                         if (is_bool($lrdd)) {
 599                                 return [];
 600                         }
 601
 602                         $path_parts = explode("/", trim(defaults($parts, 'path', ''), "/"));
 603
 604                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 605                                 $host .= "/".array_shift($path_parts);
 606                                 $lrdd = self::hostMeta($host);
 607                         }
 608                         if (!$lrdd) {
 609                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 610                                 return self::feed($uri);
 611                         }
 612                         $nick = array_pop($path_parts);
 613
 614                         // Mastodon uses a "@" as prefix for usernames in their url format
 615                         $nick = ltrim($nick, '@');
 616
 617                         $addr = $nick."@".$host;
 618                 } elseif (strstr($uri, '@')) {
 619                         // If the URI starts with "mailto:" then jump directly to the mail detection
 620                         if (strpos($uri, 'mailto:') !== false) {
 621                                 $uri = str_replace('mailto:', '', $uri);
 622                                 return self::mail($uri, $uid);
 623                         }
 624
 625                         if ($network == Protocol::MAIL) {
 626                                 return self::mail($uri, $uid);
 627                         }
 628                         // Remove "acct:" from the URI
 629                         $uri = str_replace('acct:', '', $uri);
 630
 631                         $host = substr($uri, strpos($uri, '@') + 1);
 632                         $nick = substr($uri, 0, strpos($uri, '@'));
 633
 634                         if (strpos($uri, '@twitter.com')) {
 635                                 return ["network" => Protocol::TWITTER];
 636                         }
 637                         $lrdd = self::hostMeta($host);
 638
 639                         if (is_bool($lrdd)) {
 640                                 return [];
 641                         }
 642
 643                         if (!$lrdd) {
 644                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
 645                                 return self::mail($uri, $uid);
 646                         }
 647                         $addr = $uri;
 648                 } else {
 649                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
 650                         return false;
 651                 }
 652
 653                 $webfinger = false;
 654
 655                 /// @todo Do we need the prefix "acct:" or "acct://"?
 656
 657                 foreach ($lrdd as $type => $template) {
 658                         if ($webfinger) {
 659                                 continue;
 660                         }
 661
 662                         // At first try it with the given uri
 663                         $path = str_replace('{uri}', urlencode($uri), $template);
 664                         $webfinger = self::webfinger($path, $type);
 665
 666                         // Fix possible problems with GNU Social probing to wrong scheme
 667                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 668
 669                         // We cannot be sure that the detected address was correct, so we don't use the values
 670                         if ($webfinger && ($uri != $addr)) {
 671                                 $nick = "";
 672                                 $addr = "";
 673                         }
 674
 675                         // Try webfinger with the address (user@domain.tld)
 676                         if (!$webfinger) {
 677                                 $path = str_replace('{uri}', urlencode($addr), $template);
 678                                 $webfinger = self::webfinger($path, $type);
 679                         }
 680
 681                         // Mastodon needs to have it with "acct:"
 682                         if (!$webfinger) {
 683                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 684                                 $webfinger = self::webfinger($path, $type);
 685                         }
 686                 }
 687
 688                 if (!$webfinger) {
 689                         return self::feed($uri);
 690                 }
 691
 692                 $result = false;
 693
 694                 Logger::log("Probing ".$uri, Logger::DEBUG);
 695
 696                 if (in_array($network, ["", Protocol::DFRN])) {
 697                         $result = self::dfrn($webfinger);
 698                 }
 699                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 700                         $result = self::diaspora($webfinger);
 701                 }
 702                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 703                         $result = self::ostatus($webfinger);
 704                 }
 705                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 706                         $result = self::pumpio($webfinger, $addr);
 707                 }
 708                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 709                         $result = self::feed($uri);
 710                 } else {
 711                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 712                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 713                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 714                                 $result["nick"] = $nick;
 715                         }
 716
 717                         if (empty($result["addr"]) && ($addr != "")) {
 718                                 $result["addr"] = $addr;
 719                         }
 720                 }
 721
 722                 if (empty($result["network"])) {
 723                         $result["network"] = Protocol::PHANTOM;
 724                 }
 725
 726                 if (empty($result["url"])) {
 727                         $result["url"] = $uri;
 728                 }
 729
 730                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
 731
 732                 if (empty($result["baseurl"])) {
 733                         $pos = strpos($result["url"], $host);
 734                         if ($pos) {
 735                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 736                         }
 737                 }
 738                 return $result;
 739         }
 740
 741         /**
 742          * @brief Perform a webfinger request.
 743          *
 744          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 745          *
 746          * @param string $url  Address that should be probed
 747          * @param string $type type
 748          *
 749          * @return array webfinger data
 750          * @throws HTTPException\InternalServerErrorException
 751          */
 752         private static function webfinger($url, $type)
 753         {
 754                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 755                 $redirects = 0;
 756
 757                 $curlResult = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 758                 if ($curlResult->isTimeout()) {
 759                         return false;
 760                 }
 761                 $data = $curlResult->getBody();
 762
 763                 $webfinger = json_decode($data, true);
 764                 if (is_array($webfinger)) {
 765                         if (!isset($webfinger["links"])) {
 766                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
 767                                 return false;
 768                         }
 769                         return $webfinger;
 770                 }
 771
 772                 // If it is not JSON, maybe it is XML
 773                 $xrd = XML::parseString($data, false);
 774                 if (!is_object($xrd)) {
 775                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
 776                         return false;
 777                 }
 778
 779                 $xrd_arr = XML::elementToArray($xrd);
 780                 if (!isset($xrd_arr["xrd"]["link"])) {
 781                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
 782                         return false;
 783                 }
 784
 785                 $webfinger = [];
 786
 787                 if (!empty($xrd_arr["xrd"]["subject"])) {
 788                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 789                 }
 790
 791                 if (!empty($xrd_arr["xrd"]["alias"])) {
 792                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 793                 }
 794
 795                 $webfinger["links"] = [];
 796
 797                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 798                         if (!empty($data["@attributes"])) {
 799                                 $attributes = $data["@attributes"];
 800                         } elseif ($value == "@attributes") {
 801                                 $attributes = $data;
 802                         } else {
 803                                 continue;
 804                         }
 805
 806                         $webfinger["links"][] = $attributes;
 807                 }
 808                 return $webfinger;
 809         }
 810
 811         /**
 812          * @brief Poll the Friendica specific noscrape page.
 813          *
 814          * "noscrape" is a faster alternative to fetch the data from the hcard.
 815          * This functionality was originally created for the directory.
 816          *
 817          * @param string $noscrape_url Link to the noscrape page
 818          * @param array  $data         The already fetched data
 819          *
 820          * @return array noscrape data
 821          * @throws HTTPException\InternalServerErrorException
 822          */
 823         private static function pollNoscrape($noscrape_url, $data)
 824         {
 825                 $curlResult = Network::curl($noscrape_url);
 826                 if ($curlResult->isTimeout()) {
 827                         return false;
 828                 }
 829                 $content = $curlResult->getBody();
 830                 if (!$content) {
 831                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
 832                         return false;
 833                 }
 834
 835                 $json = json_decode($content, true);
 836                 if (!is_array($json)) {
 837                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
 838                         return false;
 839                 }
 840
 841                 if (!empty($json["fn"])) {
 842                         $data["name"] = $json["fn"];
 843                 }
 844
 845                 if (!empty($json["addr"])) {
 846                         $data["addr"] = $json["addr"];
 847                 }
 848
 849                 if (!empty($json["nick"])) {
 850                         $data["nick"] = $json["nick"];
 851                 }
 852
 853                 if (!empty($json["guid"])) {
 854                         $data["guid"] = $json["guid"];
 855                 }
 856
 857                 if (!empty($json["comm"])) {
 858                         $data["community"] = $json["comm"];
 859                 }
 860
 861                 if (!empty($json["tags"])) {
 862                         $keywords = implode(" ", $json["tags"]);
 863                         if ($keywords != "") {
 864                                 $data["keywords"] = $keywords;
 865                         }
 866                 }
 867
 868                 $location = Profile::formatLocation($json);
 869                 if ($location) {
 870                         $data["location"] = $location;
 871                 }
 872
 873                 if (!empty($json["about"])) {
 874                         $data["about"] = $json["about"];
 875                 }
 876
 877                 if (!empty($json["key"])) {
 878                         $data["pubkey"] = $json["key"];
 879                 }
 880
 881                 if (!empty($json["photo"])) {
 882                         $data["photo"] = $json["photo"];
 883                 }
 884
 885                 if (!empty($json["dfrn-request"])) {
 886                         $data["request"] = $json["dfrn-request"];
 887                 }
 888
 889                 if (!empty($json["dfrn-confirm"])) {
 890                         $data["confirm"] = $json["dfrn-confirm"];
 891                 }
 892
 893                 if (!empty($json["dfrn-notify"])) {
 894                         $data["notify"] = $json["dfrn-notify"];
 895                 }
 896
 897                 if (!empty($json["dfrn-poll"])) {
 898                         $data["poll"] = $json["dfrn-poll"];
 899                 }
 900
 901                 return $data;
 902         }
 903
 904         /**
 905          * @brief Check for valid DFRN data
 906          *
 907          * @param array $data DFRN data
 908          *
 909          * @return int Number of errors
 910          */
 911         public static function validDfrn($data)
 912         {
 913                 $errors = 0;
 914                 if (!isset($data['key'])) {
 915                         $errors ++;
 916                 }
 917                 if (!isset($data['dfrn-request'])) {
 918                         $errors ++;
 919                 }
 920                 if (!isset($data['dfrn-confirm'])) {
 921                         $errors ++;
 922                 }
 923                 if (!isset($data['dfrn-notify'])) {
 924                         $errors ++;
 925                 }
 926                 if (!isset($data['dfrn-poll'])) {
 927                         $errors ++;
 928                 }
 929                 return $errors;
 930         }
 931
 932         /**
 933          * @brief Fetch data from a DFRN profile page and via "noscrape"
 934          *
 935          * @param string $profile_link Link to the profile page
 936          *
 937          * @return array profile data
 938          * @throws HTTPException\InternalServerErrorException
 939          * @throws \ImagickException
 940          */
 941         public static function profile($profile_link)
 942         {
 943                 $data = [];
 944
 945                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
 946
 947                 // Fetch data via noscrape - this is faster
 948                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 949                 $data = self::pollNoscrape($noscrape_url, $data);
 950
 951                 if (!isset($data["notify"])
 952                         || !isset($data["confirm"])
 953                         || !isset($data["request"])
 954                         || !isset($data["poll"])
 955                         || !isset($data["name"])
 956                         || !isset($data["photo"])
 957                 ) {
 958                         $data = self::pollHcard($profile_link, $data, true);
 959                 }
 960
 961                 $prof_data = [];
 962
 963                 if (empty($data["addr"]) || empty($data["nick"])) {
 964                         $probe_data = self::uri($profile_link);
 965                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 966                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 967                 }
 968
 969                 $prof_data["addr"]         = $data["addr"];
 970                 $prof_data["nick"]         = $data["nick"];
 971                 $prof_data["dfrn-request"] = defaults($data, 'request', null);
 972                 $prof_data["dfrn-confirm"] = defaults($data, 'confirm', null);
 973                 $prof_data["dfrn-notify"]  = defaults($data, 'notify' , null);
 974                 $prof_data["dfrn-poll"]    = defaults($data, 'poll'   , null);
 975                 $prof_data["photo"]        = defaults($data, 'photo'  , null);
 976                 $prof_data["fn"]           = defaults($data, 'name'   , null);
 977                 $prof_data["key"]          = defaults($data, 'pubkey' , null);
 978
 979                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
 980
 981                 return $prof_data;
 982         }
 983
 984         /**
 985          * @brief Check for DFRN contact
 986          *
 987          * @param array $webfinger Webfinger data
 988          *
 989          * @return array DFRN data
 990          * @throws HTTPException\InternalServerErrorException
 991          */
 992         private static function dfrn($webfinger)
 993         {
 994                 $hcard_url = "";
 995                 $data = [];
 996                 // The array is reversed to take into account the order of preference for same-rel links
 997                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
 998                 foreach (array_reverse($webfinger["links"]) as $link) {
 999                         if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) {
1000                                 $data["network"] = Protocol::DFRN;
1001                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1002                                 $data["poll"] = $link["href"];
1003                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1004                                 $data["url"] = $link["href"];
1005                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1006                                 $hcard_url = $link["href"];
1007                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1008                                 $data["poco"] = $link["href"];
1009                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1010                                 $data["photo"] = $link["href"];
1011                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1012                                 $data["baseurl"] = trim($link["href"], '/');
1013                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1014                                 $data["guid"] = $link["href"];
1015                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1016                                 $data["pubkey"] = base64_decode($link["href"]);
1017
1018                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1019                                 if (strstr($data["pubkey"], 'RSA ')) {
1020                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1021                                 }
1022                         }
1023                 }
1024
1025                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1026                         foreach ($webfinger["aliases"] as $alias) {
1027                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1028                                         $data["url"] = $alias;
1029                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1030                                         $data["alias"] = $alias;
1031                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1032                                         $data["addr"] = substr($alias, 5);
1033                                 }
1034                         }
1035                 }
1036
1037                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1038                         $data["addr"] = substr($webfinger["subject"], 5);
1039                 }
1040
1041                 if (!isset($data["network"]) || ($hcard_url == "")) {
1042                         return false;
1043                 }
1044
1045                 // Fetch data via noscrape - this is faster
1046                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1047                 $data = self::pollNoscrape($noscrape_url, $data);
1048
1049                 if (isset($data["notify"])
1050                         && isset($data["confirm"])
1051                         && isset($data["request"])
1052                         && isset($data["poll"])
1053                         && isset($data["name"])
1054                         && isset($data["photo"])
1055                 ) {
1056                         return $data;
1057                 }
1058
1059                 $data = self::pollHcard($hcard_url, $data, true);
1060
1061                 return $data;
1062         }
1063
1064         /**
1065          * @brief Poll the hcard page (Diaspora and Friendica specific)
1066          *
1067          * @param string  $hcard_url Link to the hcard page
1068          * @param array   $data      The already fetched data
1069          * @param boolean $dfrn      Poll DFRN specific data
1070          *
1071          * @return array hcard data
1072          * @throws HTTPException\InternalServerErrorException
1073          */
1074         private static function pollHcard($hcard_url, $data, $dfrn = false)
1075         {
1076                 $curlResult = Network::curl($hcard_url);
1077                 if ($curlResult->isTimeout()) {
1078                         return false;
1079                 }
1080                 $content = $curlResult->getBody();
1081                 if (!$content) {
1082                         return false;
1083                 }
1084
1085                 $doc = new DOMDocument();
1086                 if (!@$doc->loadHTML($content)) {
1087                         return false;
1088                 }
1089
1090                 $xpath = new DomXPath($doc);
1091
1092                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1093                 if (!is_object($vcards)) {
1094                         return false;
1095                 }
1096
1097                 if (!isset($data["baseurl"])) {
1098                         $data["baseurl"] = "";
1099                 }
1100
1101                 if ($vcards->length > 0) {
1102                         $vcard = $vcards->item(0);
1103
1104                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1105                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1106                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1107                         if (($search->length > 0) && empty($data["guid"])) {
1108                                 $data["guid"] = $search->item(0)->nodeValue;
1109                         }
1110
1111                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1112                         if ($search->length > 0) {
1113                                 $data["nick"] = $search->item(0)->nodeValue;
1114                         }
1115
1116                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1117                         if ($search->length > 0) {
1118                                 $data["name"] = $search->item(0)->nodeValue;
1119                         }
1120
1121                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1122                         if ($search->length > 0) {
1123                                 $data["searchable"] = $search->item(0)->nodeValue;
1124                         }
1125
1126                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1127                         if ($search->length > 0) {
1128                                 $data["pubkey"] = $search->item(0)->nodeValue;
1129                                 if (strstr($data["pubkey"], 'RSA ')) {
1130                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1131                                 }
1132                         }
1133
1134                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1135                         if ($search->length > 0) {
1136                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1137                         }
1138                 }
1139
1140                 $avatar = [];
1141                 if (!empty($vcard)) {
1142                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1143                         foreach ($photos as $photo) {
1144                                 $attr = [];
1145                                 foreach ($photo->attributes as $attribute) {
1146                                         $attr[$attribute->name] = trim($attribute->value);
1147                                 }
1148
1149                                 if (isset($attr["src"]) && isset($attr["width"])) {
1150                                         $avatar[$attr["width"]] = $attr["src"];
1151                                 }
1152
1153                                 // We don't have a width. So we just take everything that we got.
1154                                 // This is a Hubzilla workaround which doesn't send a width.
1155                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1156                                         $avatar[] = $attr["src"];
1157                                 }
1158                         }
1159                 }
1160
1161                 if (sizeof($avatar)) {
1162                         ksort($avatar);
1163                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1164                 }
1165
1166                 if ($dfrn) {
1167                         // Poll DFRN specific data
1168                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1169                         if ($search->length > 0) {
1170                                 foreach ($search as $link) {
1171                                         //$data["request"] = $search->item(0)->nodeValue;
1172                                         $attr = [];
1173                                         foreach ($link->attributes as $attribute) {
1174                                                 $attr[$attribute->name] = trim($attribute->value);
1175                                         }
1176
1177                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1178                                 }
1179                         }
1180
1181                         // Older Friendica versions had used the "uid" field differently than newer versions
1182                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1183                                 unset($data["guid"]);
1184                         }
1185                 }
1186
1187
1188                 return $data;
1189         }
1190
1191         /**
1192          * @brief Check for Diaspora contact
1193          *
1194          * @param array $webfinger Webfinger data
1195          *
1196          * @return array Diaspora data
1197          * @throws HTTPException\InternalServerErrorException
1198          */
1199         private static function diaspora($webfinger)
1200         {
1201                 $hcard_url = "";
1202                 $data = [];
1203                 // The array is reversed to take into account the order of preference for same-rel links
1204                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1205                 foreach (array_reverse($webfinger["links"]) as $link) {
1206                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1207                                 $hcard_url = $link["href"];
1208                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1209                                 $data["baseurl"] = trim($link["href"], '/');
1210                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1211                                 $data["guid"] = $link["href"];
1212                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1213                                 $data["url"] = $link["href"];
1214                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1215                                 $data["poll"] = $link["href"];
1216                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1217                                 $data["poco"] = $link["href"];
1218                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1219                                 $data["notify"] = $link["href"];
1220                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1221                                 $data["pubkey"] = base64_decode($link["href"]);
1222
1223                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1224                                 if (strstr($data["pubkey"], 'RSA ')) {
1225                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1226                                 }
1227                         }
1228                 }
1229
1230                 if (!isset($data["url"]) || ($hcard_url == "")) {
1231                         return false;
1232                 }
1233
1234                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1235                         foreach ($webfinger["aliases"] as $alias) {
1236                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1237                                         $data["alias"] = $alias;
1238                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1239                                         $data["addr"] = substr($alias, 5);
1240                                 }
1241                         }
1242                 }
1243
1244                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1245                         $data["addr"] = substr($webfinger["subject"], 5);
1246                 }
1247
1248                 // Fetch further information from the hcard
1249                 $data = self::pollHcard($hcard_url, $data);
1250
1251                 if (!$data) {
1252                         return false;
1253                 }
1254
1255                 if (isset($data["url"])
1256                         && isset($data["guid"])
1257                         && isset($data["baseurl"])
1258                         && isset($data["pubkey"])
1259                         && ($hcard_url != "")
1260                 ) {
1261                         $data["network"] = Protocol::DIASPORA;
1262
1263                         // The Diaspora handle must always be lowercase
1264                         if (!empty($data["addr"])) {
1265                                 $data["addr"] = strtolower($data["addr"]);
1266                         }
1267
1268                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1269                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1270                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1271                 } else {
1272                         return false;
1273                 }
1274
1275                 return $data;
1276         }
1277
1278         /**
1279          * @brief Check for OStatus contact
1280          *
1281          * @param array $webfinger Webfinger data
1282          * @param bool  $short     Short detection mode
1283          *
1284          * @return array|bool OStatus data or "false" on error or "true" on short mode
1285          * @throws HTTPException\InternalServerErrorException
1286          */
1287         private static function ostatus($webfinger, $short = false)
1288         {
1289                 $data = [];
1290
1291                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1292                         foreach ($webfinger["aliases"] as $alias) {
1293                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1294                                         $data["addr"] = str_replace('acct:', '', $alias);
1295                                 }
1296                         }
1297                 }
1298
1299                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1300                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1301                 ) {
1302                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1303                 }
1304
1305                 $pubkey = "";
1306                 if (is_array($webfinger["links"])) {
1307                         // The array is reversed to take into account the order of preference for same-rel links
1308                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1309                         foreach (array_reverse($webfinger["links"]) as $link) {
1310                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1311                                         && (defaults($link, "type", "") == "text/html")
1312                                         && ($link["href"] != "")
1313                                 ) {
1314                                         $data["url"] = $link["href"];
1315                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1316                                         $data["notify"] = $link["href"];
1317                                 } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1318                                         $data["poll"] = $link["href"];
1319                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1320                                         $pubkey = $link["href"];
1321
1322                                         if (substr($pubkey, 0, 5) === 'data:') {
1323                                                 if (strstr($pubkey, ',')) {
1324                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1325                                                 } else {
1326                                                         $pubkey = substr($pubkey, 5);
1327                                                 }
1328                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1329                                                 $curlResult = Network::curl($pubkey);
1330                                                 if ($curlResult->isTimeout()) {
1331                                                         return false;
1332                                                 }
1333                                                 $pubkey = $curlResult->getBody();
1334                                         }
1335
1336                                         $key = explode(".", $pubkey);
1337
1338                                         if (sizeof($key) >= 3) {
1339                                                 $m = Strings::base64UrlDecode($key[1]);
1340                                                 $e = Strings::base64UrlDecode($key[2]);
1341                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1342                                         }
1343                                 }
1344                         }
1345                 }
1346
1347                 if (isset($data["notify"]) && isset($data["pubkey"])
1348                         && isset($data["poll"])
1349                         && isset($data["url"])
1350                 ) {
1351                         $data["network"] = Protocol::OSTATUS;
1352                 } else {
1353                         return false;
1354                 }
1355
1356                 if ($short) {
1357                         return true;
1358                 }
1359
1360                 // Fetch all additional data from the feed
1361                 $curlResult = Network::curl($data["poll"]);
1362                 if ($curlResult->isTimeout()) {
1363                         return false;
1364                 }
1365                 $feed = $curlResult->getBody();
1366                 $dummy1 = null;
1367                 $dummy2 = null;
1368                 $dummy2 = null;
1369                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1370                 if (!$feed_data) {
1371                         return false;
1372                 }
1373
1374                 if (!empty($feed_data["header"]["author-name"])) {
1375                         $data["name"] = $feed_data["header"]["author-name"];
1376                 }
1377                 if (!empty($feed_data["header"]["author-nick"])) {
1378                         $data["nick"] = $feed_data["header"]["author-nick"];
1379                 }
1380                 if (!empty($feed_data["header"]["author-avatar"])) {
1381                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1382                 }
1383                 if (!empty($feed_data["header"]["author-id"])) {
1384                         $data["alias"] = $feed_data["header"]["author-id"];
1385                 }
1386                 if (!empty($feed_data["header"]["author-location"])) {
1387                         $data["location"] = $feed_data["header"]["author-location"];
1388                 }
1389                 if (!empty($feed_data["header"]["author-about"])) {
1390                         $data["about"] = $feed_data["header"]["author-about"];
1391                 }
1392                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1393                 // So we take the value that we just fetched, although the other one worked as well
1394                 if (!empty($feed_data["header"]["author-link"])) {
1395                         $data["url"] = $feed_data["header"]["author-link"];
1396                 }
1397
1398                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1399                         $data['url'] = $data["alias"];
1400                         $data["alias"] = '';
1401                 }
1402
1403                 /// @todo Fetch location and "about" from the feed as well
1404                 return $data;
1405         }
1406
1407         /**
1408          * @brief Fetch data from a pump.io profile page
1409          *
1410          * @param string $profile_link Link to the profile page
1411          *
1412          * @return array profile data
1413          */
1414         private static function pumpioProfileData($profile_link)
1415         {
1416                 $doc = new DOMDocument();
1417                 if (!@$doc->loadHTMLFile($profile_link)) {
1418                         return false;
1419                 }
1420
1421                 $xpath = new DomXPath($doc);
1422
1423                 $data = [];
1424
1425                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1426
1427                 if ($data["name"] == '') {
1428                         // This is ugly - but pump.io doesn't seem to know a better way for it
1429                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1430                         $pos = strpos($data["name"], chr(10));
1431                         if ($pos) {
1432                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1433                         }
1434                 }
1435
1436                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1437
1438                 if ($data["location"] == '') {
1439                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1440                 }
1441
1442                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1443
1444                 if ($data["about"] == '') {
1445                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1446                 }
1447
1448                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1449                 if (!$avatar) {
1450                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1451                 }
1452                 if ($avatar) {
1453                         foreach ($avatar->attributes as $attribute) {
1454                                 if ($attribute->name == "src") {
1455                                         $data["photo"] = trim($attribute->value);
1456                                 }
1457                         }
1458                 }
1459
1460                 return $data;
1461         }
1462
1463         /**
1464          * @brief Check for pump.io contact
1465          *
1466          * @param array $webfinger Webfinger data
1467          *
1468          * @param       $addr
1469          * @return array pump.io data
1470          */
1471         private static function pumpio($webfinger, $addr)
1472         {
1473                 $data = [];
1474                 // The array is reversed to take into account the order of preference for same-rel links
1475                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1476                 foreach (array_reverse($webfinger["links"]) as $link) {
1477                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1478                                 && (defaults($link, "type", "") == "text/html")
1479                                 && ($link["href"] != "")
1480                         ) {
1481                                 $data["url"] = $link["href"];
1482                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1483                                 $data["notify"] = $link["href"];
1484                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1485                                 $data["poll"] = $link["href"];
1486                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1487                                 $data["dialback"] = $link["href"];
1488                         }
1489                 }
1490                 if (isset($data["poll"]) && isset($data["notify"])
1491                         && isset($data["dialback"])
1492                         && isset($data["url"])
1493                 ) {
1494                         // by now we use these fields only for the network type detection
1495                         // So we unset all data that isn't used at the moment
1496                         unset($data["dialback"]);
1497
1498                         $data["network"] = Protocol::PUMPIO;
1499                 } else {
1500                         return false;
1501                 }
1502
1503                 $profile_data = self::pumpioProfileData($data["url"]);
1504
1505                 if (!$profile_data) {
1506                         return false;
1507                 }
1508
1509                 $data = array_merge($data, $profile_data);
1510
1511                 if (($addr != '') && ($data['name'] != '')) {
1512                         $name = trim(str_replace($addr, '', $data['name']));
1513                         if ($name != '') {
1514                                 $data['name'] = $name;
1515                         }
1516                 }
1517
1518                 return $data;
1519         }
1520
1521         /**
1522          * @brief Check page for feed link
1523          *
1524          * @param string $url Page link
1525          *
1526          * @return string feed link
1527          */
1528         private static function getFeedLink($url)
1529         {
1530                 $doc = new DOMDocument();
1531
1532                 if (!@$doc->loadHTMLFile($url)) {
1533                         return false;
1534                 }
1535
1536                 $xpath = new DomXPath($doc);
1537
1538                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1539                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1540                 if (!is_object($feeds)) {
1541                         return false;
1542                 }
1543
1544                 if ($feeds->length == 0) {
1545                         return false;
1546                 }
1547
1548                 $feed_url = "";
1549
1550                 foreach ($feeds as $feed) {
1551                         $attr = [];
1552                         foreach ($feed->attributes as $attribute) {
1553                                 $attr[$attribute->name] = trim($attribute->value);
1554                         }
1555
1556                         if ($feed_url == "") {
1557                                 $feed_url = $attr["href"];
1558                         }
1559                 }
1560
1561                 return $feed_url;
1562         }
1563
1564         /**
1565          * @brief Check for feed contact
1566          *
1567          * @param string  $url   Profile link
1568          * @param boolean $probe Do a probe if the page contains a feed link
1569          *
1570          * @return array feed data
1571          * @throws HTTPException\InternalServerErrorException
1572          */
1573         private static function feed($url, $probe = true)
1574         {
1575                 $curlResult = Network::curl($url);
1576                 if ($curlResult->isTimeout()) {
1577                         return false;
1578                 }
1579                 $feed = $curlResult->getBody();
1580                 $dummy1 = $dummy2 = $dummy3 = null;
1581                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1582
1583                 if (!$feed_data) {
1584                         if (!$probe) {
1585                                 return false;
1586                         }
1587
1588                         $feed_url = self::getFeedLink($url);
1589
1590                         if (!$feed_url) {
1591                                 return false;
1592                         }
1593
1594                         return self::feed($feed_url, false);
1595                 }
1596
1597                 if (!empty($feed_data["header"]["author-name"])) {
1598                         $data["name"] = $feed_data["header"]["author-name"];
1599                 }
1600
1601                 if (!empty($feed_data["header"]["author-nick"])) {
1602                         $data["nick"] = $feed_data["header"]["author-nick"];
1603                 }
1604
1605                 if (!empty($feed_data["header"]["author-avatar"])) {
1606                         $data["photo"] = $feed_data["header"]["author-avatar"];
1607                 }
1608
1609                 if (!empty($feed_data["header"]["author-id"])) {
1610                         $data["alias"] = $feed_data["header"]["author-id"];
1611                 }
1612
1613                 $data["url"] = $url;
1614                 $data["poll"] = $url;
1615
1616                 if (!empty($feed_data["header"]["author-link"])) {
1617                         $data["baseurl"] = $feed_data["header"]["author-link"];
1618                 } else {
1619                         $data["baseurl"] = $data["url"];
1620                 }
1621
1622                 $data["network"] = Protocol::FEED;
1623
1624                 return $data;
1625         }
1626
1627         /**
1628          * @brief Check for mail contact
1629          *
1630          * @param string  $uri Profile link
1631          * @param integer $uid User ID
1632          *
1633          * @return array mail data
1634          * @throws \Exception
1635          */
1636         private static function mail($uri, $uid)
1637         {
1638                 if (!Network::isEmailDomainValid($uri)) {
1639                         return false;
1640                 }
1641
1642                 if ($uid == 0) {
1643                         return false;
1644                 }
1645
1646                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1647
1648                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1649                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1650                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1651
1652                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1653                         return false;
1654                 }
1655
1656                 $mailbox = Email::constructMailboxName($mailacct);
1657                 $password = '';
1658                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1659                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1660                 if (!$mbox) {
1661                         return false;
1662                 }
1663
1664                 $msgs = Email::poll($mbox, $uri);
1665                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1666
1667                 if (!count($msgs)) {
1668                         return false;
1669                 }
1670
1671                 $phost = substr($uri, strpos($uri, '@') + 1);
1672
1673                 $data = [];
1674                 $data["addr"]    = $uri;
1675                 $data["network"] = Protocol::MAIL;
1676                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1677                 $data["nick"]    = $data["name"];
1678                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1679                 $data["url"]     = 'mailto:'.$uri;
1680                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1681                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1682
1683                 $x = Email::messageMeta($mbox, $msgs[0]);
1684                 if (stristr($x[0]->from, $uri)) {
1685                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1686                 } elseif (stristr($x[0]->to, $uri)) {
1687                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1688                 }
1689                 if (isset($adr)) {
1690                         foreach ($adr as $feadr) {
1691                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1692                                         &&(strcasecmp($feadr->host, $phost) == 0)
1693                                         && (strlen($feadr->personal))
1694                                 ) {
1695                                         $personal = imap_mime_header_decode($feadr->personal);
1696                                         $data["name"] = "";
1697                                         foreach ($personal as $perspart) {
1698                                                 if ($perspart->charset != "default") {
1699                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1700                                                 } else {
1701                                                         $data["name"] .= $perspart->text;
1702                                                 }
1703                                         }
1704
1705                                         $data["name"] = Strings::escapeTags($data["name"]);
1706                                 }
1707                         }
1708                 }
1709                 if (!empty($mbox)) {
1710                         imap_close($mbox);
1711                 }
1712                 return $data;
1713         }
1714
1715         /**
1716          * @brief Mix two paths together to possibly fix missing parts
1717          *
1718          * @param string $avatar Path to the avatar
1719          * @param string $base   Another path that is hopefully complete
1720          *
1721          * @return string fixed avatar path
1722          * @throws \Exception
1723          */
1724         public static function fixAvatar($avatar, $base)
1725         {
1726                 $base_parts = parse_url($base);
1727
1728                 // Remove all parts that could create a problem
1729                 unset($base_parts['path']);
1730                 unset($base_parts['query']);
1731                 unset($base_parts['fragment']);
1732
1733                 $avatar_parts = parse_url($avatar);
1734
1735                 // Now we mix them
1736                 $parts = array_merge($base_parts, $avatar_parts);
1737
1738                 // And put them together again
1739                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1740                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1741                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1742                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1743                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1744                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1745
1746                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1747
1748                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
1749
1750                 return $fixed;
1751         }
1752 }