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\Protocol;
  16 use Friendica\Core\System;
  17 use Friendica\Database\DBA;
  18 use Friendica\Model\Contact;
  19 use Friendica\Model\Profile;
  20 use Friendica\Protocol\Email;
  21 use Friendica\Protocol\Feed;
  22 use Friendica\Util\Crypto;
  23 use Friendica\Util\DateTimeFormat;
  24 use Friendica\Util\Network;
  25 use Friendica\Util\XML;
  26 use DomXPath;
  27
  28 require_once 'include/dba.php';
  29
  30 /**
  31  * @brief This class contain functions for probing URL
  32  *
  33  */
  34 class Probe
  35 {
  36         private static $baseurl;
  37
  38         /**
  39          * @brief Rearrange the array so that it always has the same order
  40          *
  41          * @param array $data Unordered data
  42          *
  43          * @return array Ordered data
  44          */
  45         private static function rearrangeData($data)
  46         {
  47                 $fields = ["name", "nick", "guid", "url", "addr", "alias",
  48                                 "photo", "community", "keywords", "location", "about",
  49                                 "batch", "notify", "poll", "request", "confirm", "poco",
  50                                 "priority", "network", "pubkey", "baseurl"];
  51
  52                 $newdata = [];
  53                 foreach ($fields as $field) {
  54                         if (isset($data[$field])) {
  55                                 $newdata[$field] = $data[$field];
  56                         } else {
  57                                 $newdata[$field] = "";
  58                         }
  59                 }
  60
  61                 // We don't use the "priority" field anymore and replace it with a dummy.
  62                 $newdata["priority"] = 0;
  63
  64                 return $newdata;
  65         }
  66
  67         /**
  68          * @brief Check if the hostname belongs to the own server
  69          *
  70          * @param string $host The hostname that is to be checked
  71          *
  72          * @return bool Does the testes hostname belongs to the own server?
  73          */
  74         private static function ownHost($host)
  75         {
  76                 $own_host = get_app()->get_hostname();
  77
  78                 $parts = parse_url($host);
  79
  80                 if (!isset($parts['scheme'])) {
  81                         $parts = parse_url('http://'.$host);
  82                 }
  83
  84                 if (!isset($parts['host'])) {
  85                         return false;
  86                 }
  87                 return $parts['host'] == $own_host;
  88         }
  89
  90         /**
  91          * @brief Probes for webfinger path via "host-meta"
  92          *
  93          * We have to check if the servers in the future still will offer this.
  94          * It seems as if it was dropped from the standard.
  95          *
  96          * @param string $host The host part of an url
  97          *
  98          * @return array with template and type of the webfinger template for JSON or XML
  99          */
 100         private static function hostMeta($host)
 101         {
 102                 // Reset the static variable
 103                 self::$baseurl = '';
 104
 105                 $ssl_url = "https://".$host."/.well-known/host-meta";
 106                 $url = "http://".$host."/.well-known/host-meta";
 107
 108                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 109                 $redirects = 0;
 110
 111                 logger("Probing for ".$host, LOGGER_DEBUG);
 112                 $xrd = null;
 113
 114                 $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 115                 if ($ret['success']) {
 116                         $xml = $ret['body'];
 117                         $xrd = XML::parseString($xml, false);
 118                         $host_url = 'https://'.$host;
 119                 }
 120
 121                 if (!is_object($xrd)) {
 122                         $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 123                         if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 124                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 125                                 return false;
 126                         }
 127                         $xml = $ret['body'];
 128                         $xrd = XML::parseString($xml, false);
 129                         $host_url = 'http://'.$host;
 130                 }
 131                 if (!is_object($xrd)) {
 132                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 133                         return [];
 134                 }
 135
 136                 $links = XML::elementToArray($xrd);
 137                 if (!isset($links["xrd"]["link"])) {
 138                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 139                         return [];
 140                 }
 141
 142                 $lrdd = [];
 143                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 144                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 145                 // compared to the XML webfinger. So this is commented out by now.
 146                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 147
 148                 foreach ($links["xrd"]["link"] as $value => $link) {
 149                         if (!empty($link["@attributes"])) {
 150                                 $attributes = $link["@attributes"];
 151                         } elseif ($value == "@attributes") {
 152                                 $attributes = $link;
 153                         } else {
 154                                 continue;
 155                         }
 156
 157                         if (($attributes["rel"] == "lrdd") && !empty($attributes["template"])) {
 158                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 159
 160                                 $lrdd[$type] = $attributes["template"];
 161                         }
 162                 }
 163
 164                 self::$baseurl = "http://".$host;
 165
 166                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 167
 168                 return $lrdd;
 169         }
 170
 171         /**
 172          * @brief Perform Webfinger lookup and return DFRN data
 173          *
 174          * Given an email style address, perform webfinger lookup and
 175          * return the resulting DFRN profile URL, or if no DFRN profile URL
 176          * is located, returns an OStatus subscription template (prefixed
 177          * with the string 'stat:' to identify it as on OStatus template).
 178          * If this isn't an email style address just return $webbie.
 179          * Return an empty string if email-style addresses but webfinger fails,
 180          * or if the resultant personal XRD doesn't contain a supported
 181          * subscription/friend-request attribute.
 182          *
 183          * amended 7/9/2011 to return an hcard which could save potentially loading
 184          * a lengthy content page to scrape dfrn attributes
 185          *
 186          * @param string $webbie    Address that should be probed
 187          * @param string $hcard_url Link to the hcard - is returned by reference
 188          *
 189          * @return string profile link
 190          */
 191         public static function webfingerDfrn($webbie, &$hcard_url)
 192         {
 193                 $profile_link = '';
 194
 195                 $links = self::lrdd($webbie);
 196                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 197                 if (count($links)) {
 198                         foreach ($links as $link) {
 199                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 200                                         $profile_link = $link['@attributes']['href'];
 201                                 }
 202                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 203                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 204                                 }
 205                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 206                                         $hcard_url = $link['@attributes']['href'];
 207                                 }
 208                         }
 209                 }
 210                 return $profile_link;
 211         }
 212
 213         /**
 214          * @brief Check an URI for LRDD data
 215          *
 216          * this is a replacement for the "lrdd" function.
 217          * It isn't used in this class and has some redundancies in the code.
 218          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 219          *
 220          * @param string $uri Address that should be probed
 221          *
 222          * @return array uri data
 223          */
 224         public static function lrdd($uri)
 225         {
 226                 $lrdd = self::hostMeta($uri);
 227                 $webfinger = null;
 228
 229                 if (is_bool($lrdd)) {
 230                         return [];
 231                 }
 232
 233                 if (!$lrdd) {
 234                         $parts = @parse_url($uri);
 235                         if (!$parts) {
 236                                 return [];
 237                         }
 238
 239                         $host = $parts["host"];
 240                         if (!empty($parts["port"])) {
 241                                 $host .= ':'.$parts["port"];
 242                         }
 243
 244                         $path_parts = explode("/", trim($parts["path"], "/"));
 245
 246                         $nick = array_pop($path_parts);
 247
 248                         do {
 249                                 $lrdd = self::hostMeta($host);
 250                                 $host .= "/".array_shift($path_parts);
 251                         } while (!$lrdd && (sizeof($path_parts) > 0));
 252                 }
 253
 254                 if (!$lrdd) {
 255                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 256                         return [];
 257                 }
 258
 259                 foreach ($lrdd as $type => $template) {
 260                         if ($webfinger) {
 261                                 continue;
 262                         }
 263
 264                         $path = str_replace('{uri}', urlencode($uri), $template);
 265                         $webfinger = self::webfinger($path, $type);
 266
 267                         if (!$webfinger && (strstr($uri, "@"))) {
 268                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 269                                 $webfinger = self::webfinger($path, $type);
 270                         }
 271
 272                         // Special treatment for Mastodon
 273                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 274                         // But the webfinger for this format fails.
 275                         if (!$webfinger && !empty($nick)) {
 276                                 // Mastodon uses a "@" as prefix for usernames in their url format
 277                                 $nick = ltrim($nick, '@');
 278
 279                                 $addr = $nick."@".$host;
 280
 281                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 282                                 $webfinger = self::webfinger($path, $type);
 283                         }
 284                 }
 285
 286                 if (!is_array($webfinger["links"])) {
 287                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 288                         return false;
 289                 }
 290
 291                 $data = [];
 292
 293                 foreach ($webfinger["links"] as $link) {
 294                         $data[] = ["@attributes" => $link];
 295                 }
 296
 297                 if (is_array($webfinger["aliases"])) {
 298                         foreach ($webfinger["aliases"] as $alias) {
 299                                 $data[] = ["@attributes" =>
 300                                                         ["rel" => "alias",
 301                                                                 "href" => $alias]];
 302                         }
 303                 }
 304
 305                 return $data;
 306         }
 307
 308         /**
 309          * @brief Fetch information (protocol endpoints and user information) about a given uri
 310          *
 311          * @param string  $uri     Address that should be probed
 312          * @param string  $network Test for this specific network
 313          * @param integer $uid     User ID for the probe (only used for mails)
 314          * @param boolean $cache   Use cached values?
 315          *
 316          * @return array uri data
 317          */
 318         public static function uri($uri, $network = "", $uid = -1, $cache = true)
 319         {
 320                 if ($cache) {
 321                         $result = Cache::get("Probe::uri:".$network.":".$uri);
 322                         if (!is_null($result)) {
 323                                 return $result;
 324                         }
 325                 }
 326
 327                 if ($uid == -1) {
 328                         $uid = local_user();
 329                 }
 330
 331                 $data = self::detect($uri, $network, $uid);
 332
 333                 if (!isset($data["url"])) {
 334                         $data["url"] = $uri;
 335                 }
 336
 337                 if (x($data, "photo")) {
 338                         $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
 339                 } else {
 340                         $data["photo"] = System::baseUrl().'/images/person-175.jpg';
 341                 }
 342
 343                 if (empty($data["name"])) {
 344                         if (!empty($data["nick"])) {
 345                                 $data["name"] = $data["nick"];
 346                         }
 347
 348                         if (!x($data, "name")) {
 349                                 $data["name"] = $data["url"];
 350                         }
 351                 }
 352
 353                 if (empty($data["nick"])) {
 354                         $data["nick"] = strtolower($data["name"]);
 355
 356                         if (strpos($data['nick'], ' ')) {
 357                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 358                         }
 359                 }
 360
 361                 if (!empty(self::$baseurl)) {
 362                         $data["baseurl"] = self::$baseurl;
 363                 }
 364
 365                 if (empty($data["network"])) {
 366                         $data["network"] = Protocol::PHANTOM;
 367                 }
 368
 369                 $data = self::rearrangeData($data);
 370
 371                 // Only store into the cache if the value seems to be valid
 372                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 373                         Cache::set("Probe::uri:".$network.":".$uri, $data, CACHE_DAY);
 374
 375                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 376                         /// The biggest problem is the avatar picture that could have a reduced image size.
 377                         /// It should only be updated if the existing picture isn't existing anymore.
 378                         /// We only update the contact when it is no probing for a specific network.
 379                         if (($data['network'] != Protocol::FEED)
 380                                 && ($network == "")
 381                                 && $data["name"]
 382                                 && $data["nick"]
 383                                 && $data["url"]
 384                                 && $data["addr"]
 385                                 && $data["poll"]
 386                         ) {
 387                                 $fields = ['name' => $data['name'],
 388                                                 'nick' => $data['nick'],
 389                                                 'url' => $data['url'],
 390                                                 'addr' => $data['addr'],
 391                                                 'photo' => $data['photo'],
 392                                                 'keywords' => $data['keywords'],
 393                                                 'location' => $data['location'],
 394                                                 'about' => $data['about'],
 395                                                 'notify' => $data['notify'],
 396                                                 'network' => $data['network'],
 397                                                 'server_url' => $data['baseurl']];
 398
 399                                 // This doesn't cover the case when a community isn't a community anymore
 400                                 if (!empty($data['community']) && $data['community']) {
 401                                         $fields['community'] = $data['community'];
 402                                         $fields['contact-type'] = Contact::ACCOUNT_TYPE_COMMUNITY;
 403                                 }
 404
 405                                 $fieldnames = [];
 406
 407                                 foreach ($fields as $key => $val) {
 408                                         if (empty($val)) {
 409                                                 unset($fields[$key]);
 410                                         } else {
 411                                                 $fieldnames[] = $key;
 412                                         }
 413                                 }
 414
 415                                 $fields['updated'] = DateTimeFormat::utcNow();
 416
 417                                 $condition = ['nurl' => normalise_link($data["url"])];
 418
 419                                 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
 420
 421                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 422                                 // In difference to the public contacts we want to have every contact
 423                                 // in the world in our global contacts.
 424                                 if (!$old_fields) {
 425                                         $old_fields = true;
 426
 427                                         // These values have to be set only on insert
 428                                         $fields['photo'] = $data['photo'];
 429                                         $fields['created'] = DateTimeFormat::utcNow();
 430                                 }
 431
 432                                 DBA::update('gcontact', $fields, $condition, $old_fields);
 433
 434                                 $fields = ['name' => $data['name'],
 435                                                 'nick' => $data['nick'],
 436                                                 'url' => $data['url'],
 437                                                 'addr' => $data['addr'],
 438                                                 'alias' => $data['alias'],
 439                                                 'keywords' => $data['keywords'],
 440                                                 'location' => $data['location'],
 441                                                 'about' => $data['about'],
 442                                                 'batch' => $data['batch'],
 443                                                 'notify' => $data['notify'],
 444                                                 'poll' => $data['poll'],
 445                                                 'request' => $data['request'],
 446                                                 'confirm' => $data['confirm'],
 447                                                 'poco' => $data['poco'],
 448                                                 'network' => $data['network'],
 449                                                 'pubkey' => $data['pubkey'],
 450                                                 'priority' => $data['priority'],
 451                                                 'writable' => true,
 452                                                 'rel' => Contact::SHARING];
 453
 454                                 $fieldnames = [];
 455
 456                                 foreach ($fields as $key => $val) {
 457                                         if (empty($val)) {
 458                                                 unset($fields[$key]);
 459                                         } else {
 460                                                 $fieldnames[] = $key;
 461                                         }
 462                                 }
 463
 464                                 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 465
 466                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 467                                 // This won't trigger an insert. This is intended, since we only need
 468                                 // public contacts for everyone we store items from.
 469                                 // We don't need to store every contact on the planet.
 470                                 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 471
 472                                 $fields['name-date'] = DateTimeFormat::utcNow();
 473                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 474                                 $fields['success_update'] = DateTimeFormat::utcNow();
 475
 476                                 DBA::update('contact', $fields, $condition, $old_fields);
 477                         }
 478                 }
 479
 480                 return $data;
 481         }
 482
 483         /**
 484          * @brief Switch the scheme of an url between http and https
 485          *
 486          * @param string $url URL
 487          *
 488          * @return string switched URL
 489          */
 490         private static function switchScheme($url)
 491         {
 492                 $parts = parse_url($url);
 493
 494                 if (!isset($parts['scheme'])) {
 495                         return $url;
 496                 }
 497
 498                 if ($parts['scheme'] == 'http') {
 499                         $url = str_replace('http://', 'https://', $url);
 500                 } elseif ($parts['scheme'] == 'https') {
 501                         $url = str_replace('https://', 'http://', $url);
 502                 }
 503
 504                 return $url;
 505         }
 506
 507         /**
 508          * @brief Checks if a profile url should be OStatus but only provides partial information
 509          *
 510          * @param array  $webfinger Webfinger data
 511          * @param string $lrdd      Path template for webfinger request
 512          * @param string $type      type
 513          *
 514          * @return array fixed webfinger data
 515          */
 516         private static function fixOStatus($webfinger, $lrdd, $type)
 517         {
 518                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 519                         return $webfinger;
 520                 }
 521
 522                 $is_ostatus = false;
 523                 $has_key = false;
 524
 525                 foreach ($webfinger['links'] as $link) {
 526                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 527                                 $is_ostatus = true;
 528                         }
 529                         if ($link['rel'] == 'magic-public-key') {
 530                                 $has_key = true;
 531                         }
 532                 }
 533
 534                 if (!$is_ostatus || $has_key) {
 535                         return $webfinger;
 536                 }
 537
 538                 $url = self::switchScheme($webfinger['subject']);
 539                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 540                 $webfinger2 = self::webfinger($path, $type);
 541
 542                 // Is the new webfinger detectable as OStatus?
 543                 if (self::ostatus($webfinger2, true)) {
 544                         $webfinger = $webfinger2;
 545                 }
 546
 547                 return $webfinger;
 548         }
 549
 550         /**
 551          * @brief Fetch information (protocol endpoints and user information) about a given uri
 552          *
 553          * This function is only called by the "uri" function that adds caching and rearranging of data.
 554          *
 555          * @param string  $uri     Address that should be probed
 556          * @param string  $network Test for this specific network
 557          * @param integer $uid     User ID for the probe (only used for mails)
 558          *
 559          * @return array uri data
 560          */
 561         private static function detect($uri, $network, $uid)
 562         {
 563                 $parts = parse_url($uri);
 564
 565                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 566                         $host = $parts["host"];
 567                         if (!empty($parts["port"])) {
 568                                 $host .= ':'.$parts["port"];
 569                         }
 570
 571                         if ($host == 'twitter.com') {
 572                                 return ["network" => Protocol::TWITTER];
 573                         }
 574                         $lrdd = self::hostMeta($host);
 575
 576                         if (is_bool($lrdd)) {
 577                                 return [];
 578                         }
 579
 580                         $path_parts = explode("/", trim($parts["path"], "/"));
 581
 582                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 583                                 $host .= "/".array_shift($path_parts);
 584                                 $lrdd = self::hostMeta($host);
 585                         }
 586                         if (!$lrdd) {
 587                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 588                                 return self::feed($uri);
 589                         }
 590                         $nick = array_pop($path_parts);
 591
 592                         // Mastodon uses a "@" as prefix for usernames in their url format
 593                         $nick = ltrim($nick, '@');
 594
 595                         $addr = $nick."@".$host;
 596                 } elseif (strstr($uri, '@')) {
 597                         // If the URI starts with "mailto:" then jump directly to the mail detection
 598                         if (strpos($uri, 'mailto:') !== false) {
 599                                 $uri = str_replace('mailto:', '', $uri);
 600                                 return self::mail($uri, $uid);
 601                         }
 602
 603                         if ($network == Protocol::MAIL) {
 604                                 return self::mail($uri, $uid);
 605                         }
 606                         // Remove "acct:" from the URI
 607                         $uri = str_replace('acct:', '', $uri);
 608
 609                         $host = substr($uri, strpos($uri, '@') + 1);
 610                         $nick = substr($uri, 0, strpos($uri, '@'));
 611
 612                         if (strpos($uri, '@twitter.com')) {
 613                                 return ["network" => Protocol::TWITTER];
 614                         }
 615                         $lrdd = self::hostMeta($host);
 616
 617                         if (is_bool($lrdd)) {
 618                                 return [];
 619                         }
 620
 621                         if (!$lrdd) {
 622                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 623                                 return self::mail($uri, $uid);
 624                         }
 625                         $addr = $uri;
 626                 } else {
 627                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 628                         return false;
 629                 }
 630
 631                 $webfinger = false;
 632
 633                 /// @todo Do we need the prefix "acct:" or "acct://"?
 634
 635                 foreach ($lrdd as $type => $template) {
 636                         if ($webfinger) {
 637                                 continue;
 638                         }
 639
 640                         // At first try it with the given uri
 641                         $path = str_replace('{uri}', urlencode($uri), $template);
 642                         $webfinger = self::webfinger($path, $type);
 643
 644                         // Fix possible problems with GNU Social probing to wrong scheme
 645                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 646
 647                         // We cannot be sure that the detected address was correct, so we don't use the values
 648                         if ($webfinger && ($uri != $addr)) {
 649                                 $nick = "";
 650                                 $addr = "";
 651                         }
 652
 653                         // Try webfinger with the address (user@domain.tld)
 654                         if (!$webfinger) {
 655                                 $path = str_replace('{uri}', urlencode($addr), $template);
 656                                 $webfinger = self::webfinger($path, $type);
 657                         }
 658
 659                         // Mastodon needs to have it with "acct:"
 660                         if (!$webfinger) {
 661                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 662                                 $webfinger = self::webfinger($path, $type);
 663                         }
 664                 }
 665
 666                 if (!$webfinger) {
 667                         return self::feed($uri);
 668                 }
 669
 670                 $result = false;
 671
 672                 logger("Probing ".$uri, LOGGER_DEBUG);
 673
 674                 if (in_array($network, ["", Protocol::DFRN])) {
 675                         $result = self::dfrn($webfinger);
 676                 }
 677                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 678                         $result = self::diaspora($webfinger);
 679                 }
 680                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 681                         $result = self::ostatus($webfinger);
 682                 }
 683                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 684                         $result = self::pumpio($webfinger, $addr);
 685                 }
 686                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 687                         $result = self::feed($uri);
 688                 } else {
 689                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 690                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 691                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 692                                 $result["nick"] = $nick;
 693                         }
 694
 695                         if (empty($result["addr"]) && ($addr != "")) {
 696                                 $result["addr"] = $addr;
 697                         }
 698                 }
 699
 700                 if (empty($result["network"])) {
 701                         $result["network"] = Protocol::PHANTOM;
 702                 }
 703
 704                 if (empty($result["url"])) {
 705                         $result["url"] = $uri;
 706                 }
 707
 708                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 709
 710                 if (empty($result["baseurl"])) {
 711                         $pos = strpos($result["url"], $host);
 712                         if ($pos) {
 713                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 714                         }
 715                 }
 716                 return $result;
 717         }
 718
 719         /**
 720          * @brief Perform a webfinger request.
 721          *
 722          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 723          *
 724          * @param string $url  Address that should be probed
 725          * @param string $type type
 726          *
 727          * @return array webfinger data
 728          */
 729         private static function webfinger($url, $type)
 730         {
 731                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 732                 $redirects = 0;
 733
 734                 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 735                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 736                         return false;
 737                 }
 738                 $data = $ret['body'];
 739
 740                 $webfinger = json_decode($data, true);
 741                 if (is_array($webfinger)) {
 742                         if (!isset($webfinger["links"])) {
 743                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 744                                 return false;
 745                         }
 746                         return $webfinger;
 747                 }
 748
 749                 // If it is not JSON, maybe it is XML
 750                 $xrd = XML::parseString($data, false);
 751                 if (!is_object($xrd)) {
 752                         logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
 753                         return false;
 754                 }
 755
 756                 $xrd_arr = XML::elementToArray($xrd);
 757                 if (!isset($xrd_arr["xrd"]["link"])) {
 758                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 759                         return false;
 760                 }
 761
 762                 $webfinger = [];
 763
 764                 if (!empty($xrd_arr["xrd"]["subject"])) {
 765                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 766                 }
 767
 768                 if (!empty($xrd_arr["xrd"]["alias"])) {
 769                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 770                 }
 771
 772                 $webfinger["links"] = [];
 773
 774                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 775                         if (!empty($data["@attributes"])) {
 776                                 $attributes = $data["@attributes"];
 777                         } elseif ($value == "@attributes") {
 778                                 $attributes = $data;
 779                         } else {
 780                                 continue;
 781                         }
 782
 783                         $webfinger["links"][] = $attributes;
 784                 }
 785                 return $webfinger;
 786         }
 787
 788         /**
 789          * @brief Poll the Friendica specific noscrape page.
 790          *
 791          * "noscrape" is a faster alternative to fetch the data from the hcard.
 792          * This functionality was originally created for the directory.
 793          *
 794          * @param string $noscrape_url Link to the noscrape page
 795          * @param array  $data         The already fetched data
 796          *
 797          * @return array noscrape data
 798          */
 799         private static function pollNoscrape($noscrape_url, $data)
 800         {
 801                 $ret = Network::curl($noscrape_url);
 802                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 803                         return false;
 804                 }
 805                 $content = $ret['body'];
 806                 if (!$content) {
 807                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 808                         return false;
 809                 }
 810
 811                 $json = json_decode($content, true);
 812                 if (!is_array($json)) {
 813                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 814                         return false;
 815                 }
 816
 817                 if (!empty($json["fn"])) {
 818                         $data["name"] = $json["fn"];
 819                 }
 820
 821                 if (!empty($json["addr"])) {
 822                         $data["addr"] = $json["addr"];
 823                 }
 824
 825                 if (!empty($json["nick"])) {
 826                         $data["nick"] = $json["nick"];
 827                 }
 828
 829                 if (!empty($json["guid"])) {
 830                         $data["guid"] = $json["guid"];
 831                 }
 832
 833                 if (!empty($json["comm"])) {
 834                         $data["community"] = $json["comm"];
 835                 }
 836
 837                 if (!empty($json["tags"])) {
 838                         $keywords = implode(" ", $json["tags"]);
 839                         if ($keywords != "") {
 840                                 $data["keywords"] = $keywords;
 841                         }
 842                 }
 843
 844                 $location = Profile::formatLocation($json);
 845                 if ($location) {
 846                         $data["location"] = $location;
 847                 }
 848
 849                 if (!empty($json["about"])) {
 850                         $data["about"] = $json["about"];
 851                 }
 852
 853                 if (!empty($json["key"])) {
 854                         $data["pubkey"] = $json["key"];
 855                 }
 856
 857                 if (!empty($json["photo"])) {
 858                         $data["photo"] = $json["photo"];
 859                 }
 860
 861                 if (!empty($json["dfrn-request"])) {
 862                         $data["request"] = $json["dfrn-request"];
 863                 }
 864
 865                 if (!empty($json["dfrn-confirm"])) {
 866                         $data["confirm"] = $json["dfrn-confirm"];
 867                 }
 868
 869                 if (!empty($json["dfrn-notify"])) {
 870                         $data["notify"] = $json["dfrn-notify"];
 871                 }
 872
 873                 if (!empty($json["dfrn-poll"])) {
 874                         $data["poll"] = $json["dfrn-poll"];
 875                 }
 876
 877                 return $data;
 878         }
 879
 880         /**
 881          * @brief Check for valid DFRN data
 882          *
 883          * @param array $data DFRN data
 884          *
 885          * @return int Number of errors
 886          */
 887         public static function validDfrn($data)
 888         {
 889                 $errors = 0;
 890                 if (!isset($data['key'])) {
 891                         $errors ++;
 892                 }
 893                 if (!isset($data['dfrn-request'])) {
 894                         $errors ++;
 895                 }
 896                 if (!isset($data['dfrn-confirm'])) {
 897                         $errors ++;
 898                 }
 899                 if (!isset($data['dfrn-notify'])) {
 900                         $errors ++;
 901                 }
 902                 if (!isset($data['dfrn-poll'])) {
 903                         $errors ++;
 904                 }
 905                 return $errors;
 906         }
 907
 908         /**
 909          * @brief Fetch data from a DFRN profile page and via "noscrape"
 910          *
 911          * @param string $profile_link Link to the profile page
 912          *
 913          * @return array profile data
 914          */
 915         public static function profile($profile_link)
 916         {
 917                 $data = [];
 918
 919                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 920
 921                 // Fetch data via noscrape - this is faster
 922                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 923                 $data = self::pollNoscrape($noscrape_url, $data);
 924
 925                 if (!isset($data["notify"])
 926                         || !isset($data["confirm"])
 927                         || !isset($data["request"])
 928                         || !isset($data["poll"])
 929                         || !isset($data["name"])
 930                         || !isset($data["photo"])
 931                 ) {
 932                         $data = self::pollHcard($profile_link, $data, true);
 933                 }
 934
 935                 $prof_data = [];
 936                 $prof_data["addr"]         = $data["addr"];
 937                 $prof_data["nick"]         = $data["nick"];
 938                 $prof_data["dfrn-request"] = $data["request"];
 939                 $prof_data["dfrn-confirm"] = $data["confirm"];
 940                 $prof_data["dfrn-notify"]  = $data["notify"];
 941                 $prof_data["dfrn-poll"]    = $data["poll"];
 942                 $prof_data["photo"]        = $data["photo"];
 943                 $prof_data["fn"]           = $data["name"];
 944                 $prof_data["key"]          = $data["pubkey"];
 945
 946                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 947
 948                 return $prof_data;
 949         }
 950
 951         /**
 952          * @brief Check for DFRN contact
 953          *
 954          * @param array $webfinger Webfinger data
 955          *
 956          * @return array DFRN data
 957          */
 958         private static function dfrn($webfinger)
 959         {
 960                 $hcard_url = "";
 961                 $data = [];
 962                 foreach ($webfinger["links"] as $link) {
 963                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 964                                 $data["network"] = Protocol::DFRN;
 965                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 966                                 $data["poll"] = $link["href"];
 967                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 968                                 $data["url"] = $link["href"];
 969                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 970                                 $hcard_url = $link["href"];
 971                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 972                                 $data["poco"] = $link["href"];
 973                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 974                                 $data["photo"] = $link["href"];
 975                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 976                                 $data["baseurl"] = trim($link["href"], '/');
 977                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 978                                 $data["guid"] = $link["href"];
 979                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 980                                 $data["pubkey"] = base64_decode($link["href"]);
 981
 982                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 983                                 if (strstr($data["pubkey"], 'RSA ')) {
 984                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 985                                 }
 986                         }
 987                 }
 988
 989                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 990                         foreach ($webfinger["aliases"] as $alias) {
 991                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
 992                                         $data["alias"] = $alias;
 993                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 994                                         $data["addr"] = substr($alias, 5);
 995                                 }
 996                         }
 997                 }
 998
 999                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1000                         $data["addr"] = substr($webfinger["subject"], 5);
1001                 }
1002
1003                 if (!isset($data["network"]) || ($hcard_url == "")) {
1004                         return false;
1005                 }
1006
1007                 // Fetch data via noscrape - this is faster
1008                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1009                 $data = self::pollNoscrape($noscrape_url, $data);
1010
1011                 if (isset($data["notify"])
1012                         && isset($data["confirm"])
1013                         && isset($data["request"])
1014                         && isset($data["poll"])
1015                         && isset($data["name"])
1016                         && isset($data["photo"])
1017                 ) {
1018                         return $data;
1019                 }
1020
1021                 $data = self::pollHcard($hcard_url, $data, true);
1022
1023                 return $data;
1024         }
1025
1026         /**
1027          * @brief Poll the hcard page (Diaspora and Friendica specific)
1028          *
1029          * @param string  $hcard_url Link to the hcard page
1030          * @param array   $data      The already fetched data
1031          * @param boolean $dfrn      Poll DFRN specific data
1032          *
1033          * @return array hcard data
1034          */
1035         private static function pollHcard($hcard_url, $data, $dfrn = false)
1036         {
1037                 $ret = Network::curl($hcard_url);
1038                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1039                         return false;
1040                 }
1041                 $content = $ret['body'];
1042                 if (!$content) {
1043                         return false;
1044                 }
1045
1046                 $doc = new DOMDocument();
1047                 if (!@$doc->loadHTML($content)) {
1048                         return false;
1049                 }
1050
1051                 $xpath = new DomXPath($doc);
1052
1053                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1054                 if (!is_object($vcards)) {
1055                         return false;
1056                 }
1057
1058                 if (!isset($data["baseurl"])) {
1059                         $data["baseurl"] = "";
1060                 }
1061
1062                 if ($vcards->length > 0) {
1063                         $vcard = $vcards->item(0);
1064
1065                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1066                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1067                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1068                         if (($search->length > 0) && empty($data["guid"])) {
1069                                 $data["guid"] = $search->item(0)->nodeValue;
1070                         }
1071
1072                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1073                         if ($search->length > 0) {
1074                                 $data["nick"] = $search->item(0)->nodeValue;
1075                         }
1076
1077                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1078                         if ($search->length > 0) {
1079                                 $data["name"] = $search->item(0)->nodeValue;
1080                         }
1081
1082                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1083                         if ($search->length > 0) {
1084                                 $data["searchable"] = $search->item(0)->nodeValue;
1085                         }
1086
1087                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1088                         if ($search->length > 0) {
1089                                 $data["pubkey"] = $search->item(0)->nodeValue;
1090                                 if (strstr($data["pubkey"], 'RSA ')) {
1091                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1092                                 }
1093                         }
1094
1095                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1096                         if ($search->length > 0) {
1097                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1098                         }
1099                 }
1100
1101                 $avatar = [];
1102                 if (!empty($vcard)) {
1103                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1104                         foreach ($photos as $photo) {
1105                                 $attr = [];
1106                                 foreach ($photo->attributes as $attribute) {
1107                                         $attr[$attribute->name] = trim($attribute->value);
1108                                 }
1109
1110                                 if (isset($attr["src"]) && isset($attr["width"])) {
1111                                         $avatar[$attr["width"]] = $attr["src"];
1112                                 }
1113
1114                                 // We don't have a width. So we just take everything that we got.
1115                                 // This is a Hubzilla workaround which doesn't send a width.
1116                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1117                                         $avatar[] = $attr["src"];
1118                                 }
1119                         }
1120                 }
1121
1122                 if (sizeof($avatar)) {
1123                         ksort($avatar);
1124                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1125                 }
1126
1127                 if ($dfrn) {
1128                         // Poll DFRN specific data
1129                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1130                         if ($search->length > 0) {
1131                                 foreach ($search as $link) {
1132                                         //$data["request"] = $search->item(0)->nodeValue;
1133                                         $attr = [];
1134                                         foreach ($link->attributes as $attribute) {
1135                                                 $attr[$attribute->name] = trim($attribute->value);
1136                                         }
1137
1138                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1139                                 }
1140                         }
1141
1142                         // Older Friendica versions had used the "uid" field differently than newer versions
1143                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1144                                 unset($data["guid"]);
1145                         }
1146                 }
1147
1148
1149                 return $data;
1150         }
1151
1152         /**
1153          * @brief Check for Diaspora contact
1154          *
1155          * @param array $webfinger Webfinger data
1156          *
1157          * @return array Diaspora data
1158          */
1159         private static function diaspora($webfinger)
1160         {
1161                 $hcard_url = "";
1162                 $data = [];
1163                 foreach ($webfinger["links"] as $link) {
1164                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1165                                 $hcard_url = $link["href"];
1166                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1167                                 $data["baseurl"] = trim($link["href"], '/');
1168                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1169                                 $data["guid"] = $link["href"];
1170                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1171                                 $data["url"] = $link["href"];
1172                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1173                                 $data["poll"] = $link["href"];
1174                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1175                                 $data["poco"] = $link["href"];
1176                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1177                                 $data["notify"] = $link["href"];
1178                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1179                                 $data["pubkey"] = base64_decode($link["href"]);
1180
1181                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1182                                 if (strstr($data["pubkey"], 'RSA ')) {
1183                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1184                                 }
1185                         }
1186                 }
1187
1188                 if (!isset($data["url"]) || ($hcard_url == "")) {
1189                         return false;
1190                 }
1191
1192                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1193                         foreach ($webfinger["aliases"] as $alias) {
1194                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1195                                         $data["alias"] = $alias;
1196                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1197                                         $data["addr"] = substr($alias, 5);
1198                                 }
1199                         }
1200                 }
1201
1202                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1203                         $data["addr"] = substr($webfinger["subject"], 5);
1204                 }
1205
1206                 // Fetch further information from the hcard
1207                 $data = self::pollHcard($hcard_url, $data);
1208
1209                 if (!$data) {
1210                         return false;
1211                 }
1212
1213                 if (isset($data["url"])
1214                         && isset($data["guid"])
1215                         && isset($data["baseurl"])
1216                         && isset($data["pubkey"])
1217                         && ($hcard_url != "")
1218                 ) {
1219                         $data["network"] = Protocol::DIASPORA;
1220
1221                         // The Diaspora handle must always be lowercase
1222                         if (!empty($data["addr"])) {
1223                                 $data["addr"] = strtolower($data["addr"]);
1224                         }
1225
1226                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1227                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1228                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1229                 } else {
1230                         return false;
1231                 }
1232
1233                 return $data;
1234         }
1235
1236         /**
1237          * @brief Check for OStatus contact
1238          *
1239          * @param array $webfinger Webfinger data
1240          * @param bool  $short     Short detection mode
1241          *
1242          * @return array|bool OStatus data or "false" on error or "true" on short mode
1243          */
1244         private static function ostatus($webfinger, $short = false)
1245         {
1246                 $data = [];
1247
1248                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1249                         foreach ($webfinger["aliases"] as $alias) {
1250                                 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1251                                         $data["addr"] = str_replace('acct:', '', $alias);
1252                                 }
1253                         }
1254                 }
1255
1256                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1257                         && !strstr(normalise_link($webfinger["subject"]), "http://")
1258                 ) {
1259                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1260                 }
1261
1262                 $pubkey = "";
1263                 if (is_array($webfinger["links"])) {
1264                         foreach ($webfinger["links"] as $link) {
1265                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1266                                         && ($link["type"] == "text/html")
1267                                         && ($link["href"] != "")
1268                                 ) {
1269                                         $data["url"] = $link["href"];
1270                                 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1271                                         $data["notify"] = $link["href"];
1272                                 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1273                                         $data["poll"] = $link["href"];
1274                                 } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1275                                         $pubkey = $link["href"];
1276
1277                                         if (substr($pubkey, 0, 5) === 'data:') {
1278                                                 if (strstr($pubkey, ',')) {
1279                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1280                                                 } else {
1281                                                         $pubkey = substr($pubkey, 5);
1282                                                 }
1283                                         } elseif (normalise_link($pubkey) == 'http://') {
1284                                                 $ret = Network::curl($pubkey);
1285                                                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1286                                                         return false;
1287                                                 }
1288                                                 $pubkey = $ret['body'];
1289                                         }
1290
1291                                         $key = explode(".", $pubkey);
1292
1293                                         if (sizeof($key) >= 3) {
1294                                                 $m = base64url_decode($key[1]);
1295                                                 $e = base64url_decode($key[2]);
1296                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1297                                         }
1298                                 }
1299                         }
1300                 }
1301
1302                 if (isset($data["notify"]) && isset($data["pubkey"])
1303                         && isset($data["poll"])
1304                         && isset($data["url"])
1305                 ) {
1306                         $data["network"] = Protocol::OSTATUS;
1307                 } else {
1308                         return false;
1309                 }
1310
1311                 if ($short) {
1312                         return true;
1313                 }
1314
1315                 // Fetch all additional data from the feed
1316                 $ret = Network::curl($data["poll"]);
1317                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1318                         return false;
1319                 }
1320                 $feed = $ret['body'];
1321                 $dummy1 = null;
1322                 $dummy2 = null;
1323                 $dummy2 = null;
1324                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1325                 if (!$feed_data) {
1326                         return false;
1327                 }
1328
1329                 if (!empty($feed_data["header"]["author-name"])) {
1330                         $data["name"] = $feed_data["header"]["author-name"];
1331                 }
1332                 if (!empty($feed_data["header"]["author-nick"])) {
1333                         $data["nick"] = $feed_data["header"]["author-nick"];
1334                 }
1335                 if (!empty($feed_data["header"]["author-avatar"])) {
1336                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1337                 }
1338                 if (!empty($feed_data["header"]["author-id"])) {
1339                         $data["alias"] = $feed_data["header"]["author-id"];
1340                 }
1341                 if (!empty($feed_data["header"]["author-location"])) {
1342                         $data["location"] = $feed_data["header"]["author-location"];
1343                 }
1344                 if (!empty($feed_data["header"]["author-about"])) {
1345                         $data["about"] = $feed_data["header"]["author-about"];
1346                 }
1347                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1348                 // So we take the value that we just fetched, although the other one worked as well
1349                 if (!empty($feed_data["header"]["author-link"])) {
1350                         $data["url"] = $feed_data["header"]["author-link"];
1351                 }
1352
1353                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1354                         $data['url'] = $data["alias"];
1355                         $data["alias"] = '';
1356                 }
1357
1358                 /// @todo Fetch location and "about" from the feed as well
1359                 return $data;
1360         }
1361
1362         /**
1363          * @brief Fetch data from a pump.io profile page
1364          *
1365          * @param string $profile_link Link to the profile page
1366          *
1367          * @return array profile data
1368          */
1369         private static function pumpioProfileData($profile_link)
1370         {
1371                 $doc = new DOMDocument();
1372                 if (!@$doc->loadHTMLFile($profile_link)) {
1373                         return false;
1374                 }
1375
1376                 $xpath = new DomXPath($doc);
1377
1378                 $data = [];
1379
1380                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1381
1382                 if ($data["name"] == '') {
1383                         // This is ugly - but pump.io doesn't seem to know a better way for it
1384                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1385                         $pos = strpos($data["name"], chr(10));
1386                         if ($pos) {
1387                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1388                         }
1389                 }
1390
1391                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1392
1393                 if ($data["location"] == '') {
1394                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1395                 }
1396
1397                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1398
1399                 if ($data["about"] == '') {
1400                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1401                 }
1402
1403                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1404                 if (!$avatar) {
1405                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1406                 }
1407                 if ($avatar) {
1408                         foreach ($avatar->attributes as $attribute) {
1409                                 if ($attribute->name == "src") {
1410                                         $data["photo"] = trim($attribute->value);
1411                                 }
1412                         }
1413                 }
1414
1415                 return $data;
1416         }
1417
1418         /**
1419          * @brief Check for pump.io contact
1420          *
1421          * @param array $webfinger Webfinger data
1422          *
1423          * @return array pump.io data
1424          */
1425         private static function pumpio($webfinger, $addr)
1426         {
1427                 $data = [];
1428                 foreach ($webfinger["links"] as $link) {
1429                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1430                                 && ($link["type"] == "text/html")
1431                                 && ($link["href"] != "")
1432                         ) {
1433                                 $data["url"] = $link["href"];
1434                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1435                                 $data["notify"] = $link["href"];
1436                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1437                                 $data["poll"] = $link["href"];
1438                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1439                                 $data["dialback"] = $link["href"];
1440                         }
1441                 }
1442                 if (isset($data["poll"]) && isset($data["notify"])
1443                         && isset($data["dialback"])
1444                         && isset($data["url"])
1445                 ) {
1446                         // by now we use these fields only for the network type detection
1447                         // So we unset all data that isn't used at the moment
1448                         unset($data["dialback"]);
1449
1450                         $data["network"] = Protocol::PUMPIO;
1451                 } else {
1452                         return false;
1453                 }
1454
1455                 $profile_data = self::pumpioProfileData($data["url"]);
1456
1457                 if (!$profile_data) {
1458                         return false;
1459                 }
1460
1461                 $data = array_merge($data, $profile_data);
1462
1463                 if (($addr != '') && ($data['name'] != '')) {
1464                         $name = trim(str_replace($addr, '', $data['name']));
1465                         if ($name != '') {
1466                                 $data['name'] = $name;
1467                         }
1468                 }
1469
1470                 return $data;
1471         }
1472
1473         /**
1474          * @brief Check page for feed link
1475          *
1476          * @param string $url Page link
1477          *
1478          * @return string feed link
1479          */
1480         private static function getFeedLink($url)
1481         {
1482                 $doc = new DOMDocument();
1483
1484                 if (!@$doc->loadHTMLFile($url)) {
1485                         return false;
1486                 }
1487
1488                 $xpath = new DomXPath($doc);
1489
1490                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1491                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1492                 if (!is_object($feeds)) {
1493                         return false;
1494                 }
1495
1496                 if ($feeds->length == 0) {
1497                         return false;
1498                 }
1499
1500                 $feed_url = "";
1501
1502                 foreach ($feeds as $feed) {
1503                         $attr = [];
1504                         foreach ($feed->attributes as $attribute) {
1505                                 $attr[$attribute->name] = trim($attribute->value);
1506                         }
1507
1508                         if ($feed_url == "") {
1509                                 $feed_url = $attr["href"];
1510                         }
1511                 }
1512
1513                 return $feed_url;
1514         }
1515
1516         /**
1517          * @brief Check for feed contact
1518          *
1519          * @param string  $url   Profile link
1520          * @param boolean $probe Do a probe if the page contains a feed link
1521          *
1522          * @return array feed data
1523          */
1524         private static function feed($url, $probe = true)
1525         {
1526                 $ret = Network::curl($url);
1527                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1528                         return false;
1529                 }
1530                 $feed = $ret['body'];
1531                 $dummy1 = $dummy2 = $dummy3 = null;
1532                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1533
1534                 if (!$feed_data) {
1535                         if (!$probe) {
1536                                 return false;
1537                         }
1538
1539                         $feed_url = self::getFeedLink($url);
1540
1541                         if (!$feed_url) {
1542                                 return false;
1543                         }
1544
1545                         return self::feed($feed_url, false);
1546                 }
1547
1548                 if (!empty($feed_data["header"]["author-name"])) {
1549                         $data["name"] = $feed_data["header"]["author-name"];
1550                 }
1551
1552                 if (!empty($feed_data["header"]["author-nick"])) {
1553                         $data["nick"] = $feed_data["header"]["author-nick"];
1554                 }
1555
1556                 if (!empty($feed_data["header"]["author-avatar"])) {
1557                         $data["photo"] = $feed_data["header"]["author-avatar"];
1558                 }
1559
1560                 if (!empty($feed_data["header"]["author-id"])) {
1561                         $data["alias"] = $feed_data["header"]["author-id"];
1562                 }
1563
1564                 $data["url"] = $url;
1565                 $data["poll"] = $url;
1566
1567                 if (!empty($feed_data["header"]["author-link"])) {
1568                         $data["baseurl"] = $feed_data["header"]["author-link"];
1569                 } else {
1570                         $data["baseurl"] = $data["url"];
1571                 }
1572
1573                 $data["network"] = Protocol::FEED;
1574
1575                 return $data;
1576         }
1577
1578         /**
1579          * @brief Check for mail contact
1580          *
1581          * @param string  $uri Profile link
1582          * @param integer $uid User ID
1583          *
1584          * @return array mail data
1585          */
1586         private static function mail($uri, $uid)
1587         {
1588                 if (!Network::isEmailDomainValid($uri)) {
1589                         return false;
1590                 }
1591
1592                 if ($uid == 0) {
1593                         return false;
1594                 }
1595
1596                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1597
1598                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1599
1600                 if (!DBA::isResult($x) || !DBA::isResult($r)) {
1601                         return false;
1602                 }
1603
1604                 $mailbox = Email::constructMailboxName($r[0]);
1605                 $password = '';
1606                 openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1607                 $mbox = Email::connect($mailbox, $r[0]['user'], $password);
1608                 if (!$mbox) {
1609                         return false;
1610                 }
1611
1612                 $msgs = Email::poll($mbox, $uri);
1613                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1614
1615                 if (!count($msgs)) {
1616                         return false;
1617                 }
1618
1619                 $phost = substr($uri, strpos($uri, '@') + 1);
1620
1621                 $data = [];
1622                 $data["addr"]    = $uri;
1623                 $data["network"] = Protocol::MAIL;
1624                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1625                 $data["nick"]    = $data["name"];
1626                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1627                 $data["url"]     = 'mailto:'.$uri;
1628                 $data["notify"]  = 'smtp '.random_string();
1629                 $data["poll"]    = 'email '.random_string();
1630
1631                 $x = Email::messageMeta($mbox, $msgs[0]);
1632                 if (stristr($x[0]->from, $uri)) {
1633                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1634                 } elseif (stristr($x[0]->to, $uri)) {
1635                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1636                 }
1637                 if (isset($adr)) {
1638                         foreach ($adr as $feadr) {
1639                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1640                                         &&(strcasecmp($feadr->host, $phost) == 0)
1641                                         && (strlen($feadr->personal))
1642                                 ) {
1643                                         $personal = imap_mime_header_decode($feadr->personal);
1644                                         $data["name"] = "";
1645                                         foreach ($personal as $perspart) {
1646                                                 if ($perspart->charset != "default") {
1647                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1648                                                 } else {
1649                                                         $data["name"] .= $perspart->text;
1650                                                 }
1651                                         }
1652
1653                                         $data["name"] = notags($data["name"]);
1654                                 }
1655                         }
1656                 }
1657                 if (!empty($mbox)) {
1658                         imap_close($mbox);
1659                 }
1660
1661                 return $data;
1662         }
1663
1664         /**
1665          * @brief Mix two paths together to possibly fix missing parts
1666          *
1667          * @param string $avatar Path to the avatar
1668          * @param string $base   Another path that is hopefully complete
1669          *
1670          * @return string fixed avatar path
1671          */
1672         public static function fixAvatar($avatar, $base)
1673         {
1674                 $base_parts = parse_url($base);
1675
1676                 // Remove all parts that could create a problem
1677                 unset($base_parts['path']);
1678                 unset($base_parts['query']);
1679                 unset($base_parts['fragment']);
1680
1681                 $avatar_parts = parse_url($avatar);
1682
1683                 // Now we mix them
1684                 $parts = array_merge($base_parts, $avatar_parts);
1685
1686                 // And put them together again
1687                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1688                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1689                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1690                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1691                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1692                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1693
1694                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1695
1696                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1697
1698                 return $fixed;
1699         }
1700 }