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\Protocol\ActivityPub;
  23 use Friendica\Util\Crypto;
  24 use Friendica\Util\DateTimeFormat;
  25 use Friendica\Util\Network;
  26 use Friendica\Util\XML;
  27 use DomXPath;
  28
  29 require_once 'include/dba.php';
  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()->get_hostname();
  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 $host The host part of an url
  98          *
  99          * @return array with template and type of the webfinger template for JSON or XML
 100          */
 101         private static function hostMeta($host)
 102         {
 103                 // Reset the static variable
 104                 self::$baseurl = '';
 105
 106                 $ssl_url = "https://".$host."/.well-known/host-meta";
 107                 $url = "http://".$host."/.well-known/host-meta";
 108
 109                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 110                 $redirects = 0;
 111
 112                 logger("Probing for ".$host, LOGGER_DEBUG);
 113                 $xrd = null;
 114
 115                 $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 116                 if ($ret['success']) {
 117                         $xml = $ret['body'];
 118                         $xrd = XML::parseString($xml, false);
 119                         $host_url = 'https://'.$host;
 120                 }
 121
 122                 if (!is_object($xrd)) {
 123                         $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 124                         if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 125                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 126                                 return false;
 127                         }
 128                         $xml = $ret['body'];
 129                         $xrd = XML::parseString($xml, false);
 130                         $host_url = 'http://'.$host;
 131                 }
 132                 if (!is_object($xrd)) {
 133                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 134                         return [];
 135                 }
 136
 137                 $links = XML::elementToArray($xrd);
 138                 if (!isset($links["xrd"]["link"])) {
 139                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 140                         return [];
 141                 }
 142
 143                 $lrdd = [];
 144                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 145                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 146                 // compared to the XML webfinger. So this is commented out by now.
 147                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 148
 149                 foreach ($links["xrd"]["link"] as $value => $link) {
 150                         if (!empty($link["@attributes"])) {
 151                                 $attributes = $link["@attributes"];
 152                         } elseif ($value == "@attributes") {
 153                                 $attributes = $link;
 154                         } else {
 155                                 continue;
 156                         }
 157
 158                         if (($attributes["rel"] == "lrdd") && !empty($attributes["template"])) {
 159                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 160
 161                                 $lrdd[$type] = $attributes["template"];
 162                         }
 163                 }
 164
 165                 self::$baseurl = "http://".$host;
 166
 167                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 168
 169                 return $lrdd;
 170         }
 171
 172         /**
 173          * @brief Perform Webfinger lookup and return DFRN data
 174          *
 175          * Given an email style address, perform webfinger lookup and
 176          * return the resulting DFRN profile URL, or if no DFRN profile URL
 177          * is located, returns an OStatus subscription template (prefixed
 178          * with the string 'stat:' to identify it as on OStatus template).
 179          * If this isn't an email style address just return $webbie.
 180          * Return an empty string if email-style addresses but webfinger fails,
 181          * or if the resultant personal XRD doesn't contain a supported
 182          * subscription/friend-request attribute.
 183          *
 184          * amended 7/9/2011 to return an hcard which could save potentially loading
 185          * a lengthy content page to scrape dfrn attributes
 186          *
 187          * @param string $webbie    Address that should be probed
 188          * @param string $hcard_url Link to the hcard - is returned by reference
 189          *
 190          * @return string profile link
 191          */
 192         public static function webfingerDfrn($webbie, &$hcard_url)
 193         {
 194                 $profile_link = '';
 195
 196                 $links = self::lrdd($webbie);
 197                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 198                 if (count($links)) {
 199                         foreach ($links as $link) {
 200                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 201                                         $profile_link = $link['@attributes']['href'];
 202                                 }
 203                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 204                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 205                                 }
 206                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 207                                         $hcard_url = $link['@attributes']['href'];
 208                                 }
 209                         }
 210                 }
 211                 return $profile_link;
 212         }
 213
 214         /**
 215          * @brief Check an URI for LRDD data
 216          *
 217          * this is a replacement for the "lrdd" function.
 218          * It isn't used in this class and has some redundancies in the code.
 219          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 220          *
 221          * @param string $uri Address that should be probed
 222          *
 223          * @return array uri data
 224          */
 225         public static function lrdd($uri)
 226         {
 227                 $lrdd = self::hostMeta($uri);
 228                 $webfinger = null;
 229
 230                 if (is_bool($lrdd)) {
 231                         return [];
 232                 }
 233
 234                 if (!$lrdd) {
 235                         $parts = @parse_url($uri);
 236                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 237                                 return [];
 238                         }
 239
 240                         $host = $parts["host"];
 241                         if (!empty($parts["port"])) {
 242                                 $host .= ':'.$parts["port"];
 243                         }
 244
 245                         $path_parts = explode("/", trim($parts["path"], "/"));
 246
 247                         $nick = array_pop($path_parts);
 248
 249                         do {
 250                                 $lrdd = self::hostMeta($host);
 251                                 $host .= "/".array_shift($path_parts);
 252                         } while (!$lrdd && (sizeof($path_parts) > 0));
 253                 }
 254
 255                 if (!$lrdd) {
 256                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 257                         return [];
 258                 }
 259
 260                 foreach ($lrdd as $type => $template) {
 261                         if ($webfinger) {
 262                                 continue;
 263                         }
 264
 265                         $path = str_replace('{uri}', urlencode($uri), $template);
 266                         $webfinger = self::webfinger($path, $type);
 267
 268                         if (!$webfinger && (strstr($uri, "@"))) {
 269                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 270                                 $webfinger = self::webfinger($path, $type);
 271                         }
 272
 273                         // Special treatment for Mastodon
 274                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 275                         // But the webfinger for this format fails.
 276                         if (!$webfinger && !empty($nick)) {
 277                                 // Mastodon uses a "@" as prefix for usernames in their url format
 278                                 $nick = ltrim($nick, '@');
 279
 280                                 $addr = $nick."@".$host;
 281
 282                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 283                                 $webfinger = self::webfinger($path, $type);
 284                         }
 285                 }
 286
 287                 if (!is_array($webfinger["links"])) {
 288                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 289                         return false;
 290                 }
 291
 292                 $data = [];
 293
 294                 foreach ($webfinger["links"] as $link) {
 295                         $data[] = ["@attributes" => $link];
 296                 }
 297
 298                 if (is_array($webfinger["aliases"])) {
 299                         foreach ($webfinger["aliases"] as $alias) {
 300                                 $data[] = ["@attributes" =>
 301                                                         ["rel" => "alias",
 302                                                                 "href" => $alias]];
 303                         }
 304                 }
 305
 306                 return $data;
 307         }
 308
 309         /**
 310          * @brief Fetch information (protocol endpoints and user information) about a given uri
 311          *
 312          * @param string  $uri     Address that should be probed
 313          * @param string  $network Test for this specific network
 314          * @param integer $uid     User ID for the probe (only used for mails)
 315          * @param boolean $cache   Use cached values?
 316          *
 317          * @return array uri data
 318          */
 319         public static function uri($uri, $network = "", $uid = -1, $cache = true)
 320         {
 321                 if ($cache) {
 322                         $result = Cache::get("Probe::uri:".$network.":".$uri);
 323                         if (!is_null($result)) {
 324                                 return $result;
 325                         }
 326                 }
 327
 328                 if ($uid == -1) {
 329                         $uid = local_user();
 330                 }
 331
 332                 if ($network != Protocol::ACTIVITYPUB) {
 333                         $data = self::detect($uri, $network, $uid);
 334                 }
 335
 336                 if (in_array(defaults($data, 'network', ''), ['', Protocol::PHANTOM])) {
 337                         $ap_profile = ActivityPub::fetchProfile($uri);
 338                         if (!empty($ap_profile) && ($ap_profile['network'] == Protocol::ACTIVITYPUB)) {
 339                                 $data = $ap_profile;
 340                         }
 341                 }
 342
 343                 if (!isset($data["url"])) {
 344                         $data["url"] = $uri;
 345                 }
 346
 347                 if (x($data, "photo")) {
 348                         $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
 349                 } else {
 350                         $data["photo"] = System::baseUrl().'/images/person-175.jpg';
 351                 }
 352
 353                 if (empty($data["name"])) {
 354                         if (!empty($data["nick"])) {
 355                                 $data["name"] = $data["nick"];
 356                         }
 357
 358                         if (!x($data, "name")) {
 359                                 $data["name"] = $data["url"];
 360                         }
 361                 }
 362
 363                 if (empty($data["nick"])) {
 364                         $data["nick"] = strtolower($data["name"]);
 365
 366                         if (strpos($data['nick'], ' ')) {
 367                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 368                         }
 369                 }
 370
 371                 if (!empty(self::$baseurl)) {
 372                         $data["baseurl"] = self::$baseurl;
 373                 }
 374
 375                 if (empty($data["network"])) {
 376                         $data["network"] = Protocol::PHANTOM;
 377                 }
 378
 379                 $data = self::rearrangeData($data);
 380
 381                 // Only store into the cache if the value seems to be valid
 382                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 383                         Cache::set("Probe::uri:".$network.":".$uri, $data, CACHE_DAY);
 384
 385                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 386                         /// The biggest problem is the avatar picture that could have a reduced image size.
 387                         /// It should only be updated if the existing picture isn't existing anymore.
 388                         /// We only update the contact when it is no probing for a specific network.
 389                         if (($data['network'] != Protocol::FEED)
 390                                 && ($network == "")
 391                                 && $data["name"]
 392                                 && $data["nick"]
 393                                 && $data["url"]
 394                                 && $data["addr"]
 395                                 && $data["poll"]
 396                         ) {
 397                                 $fields = ['name' => $data['name'],
 398                                                 'nick' => $data['nick'],
 399                                                 'url' => $data['url'],
 400                                                 'addr' => $data['addr'],
 401                                                 'photo' => $data['photo'],
 402                                                 'keywords' => $data['keywords'],
 403                                                 'location' => $data['location'],
 404                                                 'about' => $data['about'],
 405                                                 'notify' => $data['notify'],
 406                                                 'network' => $data['network'],
 407                                                 'server_url' => $data['baseurl']];
 408
 409                                 // This doesn't cover the case when a community isn't a community anymore
 410                                 if (!empty($data['community']) && $data['community']) {
 411                                         $fields['community'] = $data['community'];
 412                                         $fields['contact-type'] = Contact::ACCOUNT_TYPE_COMMUNITY;
 413                                 }
 414
 415                                 $fieldnames = [];
 416
 417                                 foreach ($fields as $key => $val) {
 418                                         if (empty($val)) {
 419                                                 unset($fields[$key]);
 420                                         } else {
 421                                                 $fieldnames[] = $key;
 422                                         }
 423                                 }
 424
 425                                 $fields['updated'] = DateTimeFormat::utcNow();
 426
 427                                 $condition = ['nurl' => normalise_link($data["url"])];
 428
 429                                 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
 430
 431                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 432                                 // In difference to the public contacts we want to have every contact
 433                                 // in the world in our global contacts.
 434                                 if (!$old_fields) {
 435                                         $old_fields = true;
 436
 437                                         // These values have to be set only on insert
 438                                         $fields['photo'] = $data['photo'];
 439                                         $fields['created'] = DateTimeFormat::utcNow();
 440                                 }
 441
 442                                 DBA::update('gcontact', $fields, $condition, $old_fields);
 443
 444                                 $fields = ['name' => $data['name'],
 445                                                 'nick' => $data['nick'],
 446                                                 'url' => $data['url'],
 447                                                 'addr' => $data['addr'],
 448                                                 'alias' => $data['alias'],
 449                                                 'keywords' => $data['keywords'],
 450                                                 'location' => $data['location'],
 451                                                 'about' => $data['about'],
 452                                                 'batch' => $data['batch'],
 453                                                 'notify' => $data['notify'],
 454                                                 'poll' => $data['poll'],
 455                                                 'request' => $data['request'],
 456                                                 'confirm' => $data['confirm'],
 457                                                 'poco' => $data['poco'],
 458                                                 'network' => $data['network'],
 459                                                 'pubkey' => $data['pubkey'],
 460                                                 'priority' => $data['priority'],
 461                                                 'writable' => true,
 462                                                 'rel' => Contact::SHARING];
 463
 464                                 $fieldnames = [];
 465
 466                                 foreach ($fields as $key => $val) {
 467                                         if (empty($val)) {
 468                                                 unset($fields[$key]);
 469                                         } else {
 470                                                 $fieldnames[] = $key;
 471                                         }
 472                                 }
 473
 474                                 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 475
 476                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 477                                 // This won't trigger an insert. This is intended, since we only need
 478                                 // public contacts for everyone we store items from.
 479                                 // We don't need to store every contact on the planet.
 480                                 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 481
 482                                 $fields['name-date'] = DateTimeFormat::utcNow();
 483                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 484                                 $fields['success_update'] = DateTimeFormat::utcNow();
 485
 486                                 DBA::update('contact', $fields, $condition, $old_fields);
 487                         }
 488                 }
 489
 490                 return $data;
 491         }
 492
 493         /**
 494          * @brief Switch the scheme of an url between http and https
 495          *
 496          * @param string $url URL
 497          *
 498          * @return string switched URL
 499          */
 500         private static function switchScheme($url)
 501         {
 502                 $parts = parse_url($url);
 503
 504                 if (!isset($parts['scheme'])) {
 505                         return $url;
 506                 }
 507
 508                 if ($parts['scheme'] == 'http') {
 509                         $url = str_replace('http://', 'https://', $url);
 510                 } elseif ($parts['scheme'] == 'https') {
 511                         $url = str_replace('https://', 'http://', $url);
 512                 }
 513
 514                 return $url;
 515         }
 516
 517         /**
 518          * @brief Checks if a profile url should be OStatus but only provides partial information
 519          *
 520          * @param array  $webfinger Webfinger data
 521          * @param string $lrdd      Path template for webfinger request
 522          * @param string $type      type
 523          *
 524          * @return array fixed webfinger data
 525          */
 526         private static function fixOStatus($webfinger, $lrdd, $type)
 527         {
 528                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 529                         return $webfinger;
 530                 }
 531
 532                 $is_ostatus = false;
 533                 $has_key = false;
 534
 535                 foreach ($webfinger['links'] as $link) {
 536                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 537                                 $is_ostatus = true;
 538                         }
 539                         if ($link['rel'] == 'magic-public-key') {
 540                                 $has_key = true;
 541                         }
 542                 }
 543
 544                 if (!$is_ostatus || $has_key) {
 545                         return $webfinger;
 546                 }
 547
 548                 $url = self::switchScheme($webfinger['subject']);
 549                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 550                 $webfinger2 = self::webfinger($path, $type);
 551
 552                 // Is the new webfinger detectable as OStatus?
 553                 if (self::ostatus($webfinger2, true)) {
 554                         $webfinger = $webfinger2;
 555                 }
 556
 557                 return $webfinger;
 558         }
 559
 560         /**
 561          * @brief Fetch information (protocol endpoints and user information) about a given uri
 562          *
 563          * This function is only called by the "uri" function that adds caching and rearranging of data.
 564          *
 565          * @param string  $uri     Address that should be probed
 566          * @param string  $network Test for this specific network
 567          * @param integer $uid     User ID for the probe (only used for mails)
 568          *
 569          * @return array uri data
 570          */
 571         private static function detect($uri, $network, $uid)
 572         {
 573                 $parts = parse_url($uri);
 574
 575                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 576                         $host = $parts["host"];
 577                         if (!empty($parts["port"])) {
 578                                 $host .= ':'.$parts["port"];
 579                         }
 580
 581                         if ($host == 'twitter.com') {
 582                                 return ["network" => Protocol::TWITTER];
 583                         }
 584                         $lrdd = self::hostMeta($host);
 585
 586                         if (is_bool($lrdd)) {
 587                                 return [];
 588                         }
 589
 590                         $path_parts = explode("/", trim($parts["path"], "/"));
 591
 592                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 593                                 $host .= "/".array_shift($path_parts);
 594                                 $lrdd = self::hostMeta($host);
 595                         }
 596                         if (!$lrdd) {
 597                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 598                                 return self::feed($uri);
 599                         }
 600                         $nick = array_pop($path_parts);
 601
 602                         // Mastodon uses a "@" as prefix for usernames in their url format
 603                         $nick = ltrim($nick, '@');
 604
 605                         $addr = $nick."@".$host;
 606                 } elseif (strstr($uri, '@')) {
 607                         // If the URI starts with "mailto:" then jump directly to the mail detection
 608                         if (strpos($uri, 'mailto:') !== false) {
 609                                 $uri = str_replace('mailto:', '', $uri);
 610                                 return self::mail($uri, $uid);
 611                         }
 612
 613                         if ($network == Protocol::MAIL) {
 614                                 return self::mail($uri, $uid);
 615                         }
 616                         // Remove "acct:" from the URI
 617                         $uri = str_replace('acct:', '', $uri);
 618
 619                         $host = substr($uri, strpos($uri, '@') + 1);
 620                         $nick = substr($uri, 0, strpos($uri, '@'));
 621
 622                         if (strpos($uri, '@twitter.com')) {
 623                                 return ["network" => Protocol::TWITTER];
 624                         }
 625                         $lrdd = self::hostMeta($host);
 626
 627                         if (is_bool($lrdd)) {
 628                                 return [];
 629                         }
 630
 631                         if (!$lrdd) {
 632                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 633                                 return self::mail($uri, $uid);
 634                         }
 635                         $addr = $uri;
 636                 } else {
 637                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 638                         return false;
 639                 }
 640
 641                 $webfinger = false;
 642
 643                 /// @todo Do we need the prefix "acct:" or "acct://"?
 644
 645                 foreach ($lrdd as $type => $template) {
 646                         if ($webfinger) {
 647                                 continue;
 648                         }
 649
 650                         // At first try it with the given uri
 651                         $path = str_replace('{uri}', urlencode($uri), $template);
 652                         $webfinger = self::webfinger($path, $type);
 653
 654                         // Fix possible problems with GNU Social probing to wrong scheme
 655                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 656
 657                         // We cannot be sure that the detected address was correct, so we don't use the values
 658                         if ($webfinger && ($uri != $addr)) {
 659                                 $nick = "";
 660                                 $addr = "";
 661                         }
 662
 663                         // Try webfinger with the address (user@domain.tld)
 664                         if (!$webfinger) {
 665                                 $path = str_replace('{uri}', urlencode($addr), $template);
 666                                 $webfinger = self::webfinger($path, $type);
 667                         }
 668
 669                         // Mastodon needs to have it with "acct:"
 670                         if (!$webfinger) {
 671                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 672                                 $webfinger = self::webfinger($path, $type);
 673                         }
 674                 }
 675
 676                 if (!$webfinger) {
 677                         return self::feed($uri);
 678                 }
 679
 680                 $result = false;
 681
 682                 logger("Probing ".$uri, LOGGER_DEBUG);
 683
 684                 if (in_array($network, ["", Protocol::DFRN])) {
 685                         $result = self::dfrn($webfinger);
 686                 }
 687                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 688                         $result = self::diaspora($webfinger);
 689                 }
 690                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 691                         $result = self::ostatus($webfinger);
 692                 }
 693                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 694                         $result = self::pumpio($webfinger, $addr);
 695                 }
 696                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 697                         $result = self::feed($uri);
 698                 } else {
 699                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 700                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 701                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 702                                 $result["nick"] = $nick;
 703                         }
 704
 705                         if (empty($result["addr"]) && ($addr != "")) {
 706                                 $result["addr"] = $addr;
 707                         }
 708                 }
 709
 710                 if (empty($result["network"])) {
 711                         $result["network"] = Protocol::PHANTOM;
 712                 }
 713
 714                 if (empty($result["url"])) {
 715                         $result["url"] = $uri;
 716                 }
 717
 718                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 719
 720                 if (empty($result["baseurl"])) {
 721                         $pos = strpos($result["url"], $host);
 722                         if ($pos) {
 723                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 724                         }
 725                 }
 726                 return $result;
 727         }
 728
 729         /**
 730          * @brief Perform a webfinger request.
 731          *
 732          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 733          *
 734          * @param string $url  Address that should be probed
 735          * @param string $type type
 736          *
 737          * @return array webfinger data
 738          */
 739         private static function webfinger($url, $type)
 740         {
 741                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 742                 $redirects = 0;
 743
 744                 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 745                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 746                         return false;
 747                 }
 748                 $data = $ret['body'];
 749
 750                 $webfinger = json_decode($data, true);
 751                 if (is_array($webfinger)) {
 752                         if (!isset($webfinger["links"])) {
 753                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 754                                 return false;
 755                         }
 756                         return $webfinger;
 757                 }
 758
 759                 // If it is not JSON, maybe it is XML
 760                 $xrd = XML::parseString($data, false);
 761                 if (!is_object($xrd)) {
 762                         logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
 763                         return false;
 764                 }
 765
 766                 $xrd_arr = XML::elementToArray($xrd);
 767                 if (!isset($xrd_arr["xrd"]["link"])) {
 768                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 769                         return false;
 770                 }
 771
 772                 $webfinger = [];
 773
 774                 if (!empty($xrd_arr["xrd"]["subject"])) {
 775                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 776                 }
 777
 778                 if (!empty($xrd_arr["xrd"]["alias"])) {
 779                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 780                 }
 781
 782                 $webfinger["links"] = [];
 783
 784                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 785                         if (!empty($data["@attributes"])) {
 786                                 $attributes = $data["@attributes"];
 787                         } elseif ($value == "@attributes") {
 788                                 $attributes = $data;
 789                         } else {
 790                                 continue;
 791                         }
 792
 793                         $webfinger["links"][] = $attributes;
 794                 }
 795                 return $webfinger;
 796         }
 797
 798         /**
 799          * @brief Poll the Friendica specific noscrape page.
 800          *
 801          * "noscrape" is a faster alternative to fetch the data from the hcard.
 802          * This functionality was originally created for the directory.
 803          *
 804          * @param string $noscrape_url Link to the noscrape page
 805          * @param array  $data         The already fetched data
 806          *
 807          * @return array noscrape data
 808          */
 809         private static function pollNoscrape($noscrape_url, $data)
 810         {
 811                 $ret = Network::curl($noscrape_url);
 812                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 813                         return false;
 814                 }
 815                 $content = $ret['body'];
 816                 if (!$content) {
 817                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 818                         return false;
 819                 }
 820
 821                 $json = json_decode($content, true);
 822                 if (!is_array($json)) {
 823                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 824                         return false;
 825                 }
 826
 827                 if (!empty($json["fn"])) {
 828                         $data["name"] = $json["fn"];
 829                 }
 830
 831                 if (!empty($json["addr"])) {
 832                         $data["addr"] = $json["addr"];
 833                 }
 834
 835                 if (!empty($json["nick"])) {
 836                         $data["nick"] = $json["nick"];
 837                 }
 838
 839                 if (!empty($json["guid"])) {
 840                         $data["guid"] = $json["guid"];
 841                 }
 842
 843                 if (!empty($json["comm"])) {
 844                         $data["community"] = $json["comm"];
 845                 }
 846
 847                 if (!empty($json["tags"])) {
 848                         $keywords = implode(" ", $json["tags"]);
 849                         if ($keywords != "") {
 850                                 $data["keywords"] = $keywords;
 851                         }
 852                 }
 853
 854                 $location = Profile::formatLocation($json);
 855                 if ($location) {
 856                         $data["location"] = $location;
 857                 }
 858
 859                 if (!empty($json["about"])) {
 860                         $data["about"] = $json["about"];
 861                 }
 862
 863                 if (!empty($json["key"])) {
 864                         $data["pubkey"] = $json["key"];
 865                 }
 866
 867                 if (!empty($json["photo"])) {
 868                         $data["photo"] = $json["photo"];
 869                 }
 870
 871                 if (!empty($json["dfrn-request"])) {
 872                         $data["request"] = $json["dfrn-request"];
 873                 }
 874
 875                 if (!empty($json["dfrn-confirm"])) {
 876                         $data["confirm"] = $json["dfrn-confirm"];
 877                 }
 878
 879                 if (!empty($json["dfrn-notify"])) {
 880                         $data["notify"] = $json["dfrn-notify"];
 881                 }
 882
 883                 if (!empty($json["dfrn-poll"])) {
 884                         $data["poll"] = $json["dfrn-poll"];
 885                 }
 886
 887                 return $data;
 888         }
 889
 890         /**
 891          * @brief Check for valid DFRN data
 892          *
 893          * @param array $data DFRN data
 894          *
 895          * @return int Number of errors
 896          */
 897         public static function validDfrn($data)
 898         {
 899                 $errors = 0;
 900                 if (!isset($data['key'])) {
 901                         $errors ++;
 902                 }
 903                 if (!isset($data['dfrn-request'])) {
 904                         $errors ++;
 905                 }
 906                 if (!isset($data['dfrn-confirm'])) {
 907                         $errors ++;
 908                 }
 909                 if (!isset($data['dfrn-notify'])) {
 910                         $errors ++;
 911                 }
 912                 if (!isset($data['dfrn-poll'])) {
 913                         $errors ++;
 914                 }
 915                 return $errors;
 916         }
 917
 918         /**
 919          * @brief Fetch data from a DFRN profile page and via "noscrape"
 920          *
 921          * @param string $profile_link Link to the profile page
 922          *
 923          * @return array profile data
 924          */
 925         public static function profile($profile_link)
 926         {
 927                 $data = [];
 928
 929                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 930
 931                 // Fetch data via noscrape - this is faster
 932                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 933                 $data = self::pollNoscrape($noscrape_url, $data);
 934
 935                 if (!isset($data["notify"])
 936                         || !isset($data["confirm"])
 937                         || !isset($data["request"])
 938                         || !isset($data["poll"])
 939                         || !isset($data["name"])
 940                         || !isset($data["photo"])
 941                 ) {
 942                         $data = self::pollHcard($profile_link, $data, true);
 943                 }
 944
 945                 $prof_data = [];
 946
 947                 if (empty($data["addr"]) || empty($data["nick"])) {
 948                         $probe_data = self::uri($profile_link);
 949                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 950                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 951                 }
 952
 953                 $prof_data["addr"]         = $data["addr"];
 954                 $prof_data["nick"]         = $data["nick"];
 955                 $prof_data["dfrn-request"] = $data["request"];
 956                 $prof_data["dfrn-confirm"] = $data["confirm"];
 957                 $prof_data["dfrn-notify"]  = $data["notify"];
 958                 $prof_data["dfrn-poll"]    = $data["poll"];
 959                 $prof_data["photo"]        = $data["photo"];
 960                 $prof_data["fn"]           = $data["name"];
 961                 $prof_data["key"]          = $data["pubkey"];
 962
 963                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 964
 965                 return $prof_data;
 966         }
 967
 968         /**
 969          * @brief Check for DFRN contact
 970          *
 971          * @param array $webfinger Webfinger data
 972          *
 973          * @return array DFRN data
 974          */
 975         private static function dfrn($webfinger)
 976         {
 977                 $hcard_url = "";
 978                 $data = [];
 979                 foreach ($webfinger["links"] as $link) {
 980                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 981                                 $data["network"] = Protocol::DFRN;
 982                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 983                                 $data["poll"] = $link["href"];
 984                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 985                                 $data["url"] = $link["href"];
 986                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 987                                 $hcard_url = $link["href"];
 988                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 989                                 $data["poco"] = $link["href"];
 990                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 991                                 $data["photo"] = $link["href"];
 992                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 993                                 $data["baseurl"] = trim($link["href"], '/');
 994                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 995                                 $data["guid"] = $link["href"];
 996                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 997                                 $data["pubkey"] = base64_decode($link["href"]);
 998
 999                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1000                                 if (strstr($data["pubkey"], 'RSA ')) {
1001                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1002                                 }
1003                         }
1004                 }
1005
1006                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1007                         foreach ($webfinger["aliases"] as $alias) {
1008                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1009                                         $data["url"] = $alias;
1010                                 } elseif (!strstr($alias, "@") && normalise_link($alias) != normalise_link($data["url"])) {
1011                                         $data["alias"] = $alias;
1012                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1013                                         $data["addr"] = substr($alias, 5);
1014                                 }
1015                         }
1016                 }
1017
1018                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1019                         $data["addr"] = substr($webfinger["subject"], 5);
1020                 }
1021
1022                 if (!isset($data["network"]) || ($hcard_url == "")) {
1023                         return false;
1024                 }
1025
1026                 // Fetch data via noscrape - this is faster
1027                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1028                 $data = self::pollNoscrape($noscrape_url, $data);
1029
1030                 if (isset($data["notify"])
1031                         && isset($data["confirm"])
1032                         && isset($data["request"])
1033                         && isset($data["poll"])
1034                         && isset($data["name"])
1035                         && isset($data["photo"])
1036                 ) {
1037                         return $data;
1038                 }
1039
1040                 $data = self::pollHcard($hcard_url, $data, true);
1041
1042                 return $data;
1043         }
1044
1045         /**
1046          * @brief Poll the hcard page (Diaspora and Friendica specific)
1047          *
1048          * @param string  $hcard_url Link to the hcard page
1049          * @param array   $data      The already fetched data
1050          * @param boolean $dfrn      Poll DFRN specific data
1051          *
1052          * @return array hcard data
1053          */
1054         private static function pollHcard($hcard_url, $data, $dfrn = false)
1055         {
1056                 $ret = Network::curl($hcard_url);
1057                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1058                         return false;
1059                 }
1060                 $content = $ret['body'];
1061                 if (!$content) {
1062                         return false;
1063                 }
1064
1065                 $doc = new DOMDocument();
1066                 if (!@$doc->loadHTML($content)) {
1067                         return false;
1068                 }
1069
1070                 $xpath = new DomXPath($doc);
1071
1072                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1073                 if (!is_object($vcards)) {
1074                         return false;
1075                 }
1076
1077                 if (!isset($data["baseurl"])) {
1078                         $data["baseurl"] = "";
1079                 }
1080
1081                 if ($vcards->length > 0) {
1082                         $vcard = $vcards->item(0);
1083
1084                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1085                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1086                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1087                         if (($search->length > 0) && empty($data["guid"])) {
1088                                 $data["guid"] = $search->item(0)->nodeValue;
1089                         }
1090
1091                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1092                         if ($search->length > 0) {
1093                                 $data["nick"] = $search->item(0)->nodeValue;
1094                         }
1095
1096                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1097                         if ($search->length > 0) {
1098                                 $data["name"] = $search->item(0)->nodeValue;
1099                         }
1100
1101                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1102                         if ($search->length > 0) {
1103                                 $data["searchable"] = $search->item(0)->nodeValue;
1104                         }
1105
1106                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1107                         if ($search->length > 0) {
1108                                 $data["pubkey"] = $search->item(0)->nodeValue;
1109                                 if (strstr($data["pubkey"], 'RSA ')) {
1110                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1111                                 }
1112                         }
1113
1114                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1115                         if ($search->length > 0) {
1116                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1117                         }
1118                 }
1119
1120                 $avatar = [];
1121                 if (!empty($vcard)) {
1122                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1123                         foreach ($photos as $photo) {
1124                                 $attr = [];
1125                                 foreach ($photo->attributes as $attribute) {
1126                                         $attr[$attribute->name] = trim($attribute->value);
1127                                 }
1128
1129                                 if (isset($attr["src"]) && isset($attr["width"])) {
1130                                         $avatar[$attr["width"]] = $attr["src"];
1131                                 }
1132
1133                                 // We don't have a width. So we just take everything that we got.
1134                                 // This is a Hubzilla workaround which doesn't send a width.
1135                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1136                                         $avatar[] = $attr["src"];
1137                                 }
1138                         }
1139                 }
1140
1141                 if (sizeof($avatar)) {
1142                         ksort($avatar);
1143                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1144                 }
1145
1146                 if ($dfrn) {
1147                         // Poll DFRN specific data
1148                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1149                         if ($search->length > 0) {
1150                                 foreach ($search as $link) {
1151                                         //$data["request"] = $search->item(0)->nodeValue;
1152                                         $attr = [];
1153                                         foreach ($link->attributes as $attribute) {
1154                                                 $attr[$attribute->name] = trim($attribute->value);
1155                                         }
1156
1157                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1158                                 }
1159                         }
1160
1161                         // Older Friendica versions had used the "uid" field differently than newer versions
1162                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1163                                 unset($data["guid"]);
1164                         }
1165                 }
1166
1167
1168                 return $data;
1169         }
1170
1171         /**
1172          * @brief Check for Diaspora contact
1173          *
1174          * @param array $webfinger Webfinger data
1175          *
1176          * @return array Diaspora data
1177          */
1178         private static function diaspora($webfinger)
1179         {
1180                 $hcard_url = "";
1181                 $data = [];
1182                 foreach ($webfinger["links"] as $link) {
1183                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1184                                 $hcard_url = $link["href"];
1185                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1186                                 $data["baseurl"] = trim($link["href"], '/');
1187                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1188                                 $data["guid"] = $link["href"];
1189                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1190                                 $data["url"] = $link["href"];
1191                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1192                                 $data["poll"] = $link["href"];
1193                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1194                                 $data["poco"] = $link["href"];
1195                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1196                                 $data["notify"] = $link["href"];
1197                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1198                                 $data["pubkey"] = base64_decode($link["href"]);
1199
1200                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1201                                 if (strstr($data["pubkey"], 'RSA ')) {
1202                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1203                                 }
1204                         }
1205                 }
1206
1207                 if (!isset($data["url"]) || ($hcard_url == "")) {
1208                         return false;
1209                 }
1210
1211                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1212                         foreach ($webfinger["aliases"] as $alias) {
1213                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1214                                         $data["alias"] = $alias;
1215                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1216                                         $data["addr"] = substr($alias, 5);
1217                                 }
1218                         }
1219                 }
1220
1221                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1222                         $data["addr"] = substr($webfinger["subject"], 5);
1223                 }
1224
1225                 // Fetch further information from the hcard
1226                 $data = self::pollHcard($hcard_url, $data);
1227
1228                 if (!$data) {
1229                         return false;
1230                 }
1231
1232                 if (isset($data["url"])
1233                         && isset($data["guid"])
1234                         && isset($data["baseurl"])
1235                         && isset($data["pubkey"])
1236                         && ($hcard_url != "")
1237                 ) {
1238                         $data["network"] = Protocol::DIASPORA;
1239
1240                         // The Diaspora handle must always be lowercase
1241                         if (!empty($data["addr"])) {
1242                                 $data["addr"] = strtolower($data["addr"]);
1243                         }
1244
1245                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1246                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1247                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1248                 } else {
1249                         return false;
1250                 }
1251
1252                 return $data;
1253         }
1254
1255         /**
1256          * @brief Check for OStatus contact
1257          *
1258          * @param array $webfinger Webfinger data
1259          * @param bool  $short     Short detection mode
1260          *
1261          * @return array|bool OStatus data or "false" on error or "true" on short mode
1262          */
1263         private static function ostatus($webfinger, $short = false)
1264         {
1265                 $data = [];
1266
1267                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1268                         foreach ($webfinger["aliases"] as $alias) {
1269                                 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1270                                         $data["addr"] = str_replace('acct:', '', $alias);
1271                                 }
1272                         }
1273                 }
1274
1275                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1276                         && !strstr(normalise_link($webfinger["subject"]), "http://")
1277                 ) {
1278                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1279                 }
1280
1281                 $pubkey = "";
1282                 if (is_array($webfinger["links"])) {
1283                         foreach ($webfinger["links"] as $link) {
1284                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1285                                         && ($link["type"] == "text/html")
1286                                         && ($link["href"] != "")
1287                                 ) {
1288                                         $data["url"] = $link["href"];
1289                                 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1290                                         $data["notify"] = $link["href"];
1291                                 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1292                                         $data["poll"] = $link["href"];
1293                                 } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1294                                         $pubkey = $link["href"];
1295
1296                                         if (substr($pubkey, 0, 5) === 'data:') {
1297                                                 if (strstr($pubkey, ',')) {
1298                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1299                                                 } else {
1300                                                         $pubkey = substr($pubkey, 5);
1301                                                 }
1302                                         } elseif (normalise_link($pubkey) == 'http://') {
1303                                                 $ret = Network::curl($pubkey);
1304                                                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1305                                                         return false;
1306                                                 }
1307                                                 $pubkey = $ret['body'];
1308                                         }
1309
1310                                         $key = explode(".", $pubkey);
1311
1312                                         if (sizeof($key) >= 3) {
1313                                                 $m = base64url_decode($key[1]);
1314                                                 $e = base64url_decode($key[2]);
1315                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1316                                         }
1317                                 }
1318                         }
1319                 }
1320
1321                 if (isset($data["notify"]) && isset($data["pubkey"])
1322                         && isset($data["poll"])
1323                         && isset($data["url"])
1324                 ) {
1325                         $data["network"] = Protocol::OSTATUS;
1326                 } else {
1327                         return false;
1328                 }
1329
1330                 if ($short) {
1331                         return true;
1332                 }
1333
1334                 // Fetch all additional data from the feed
1335                 $ret = Network::curl($data["poll"]);
1336                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1337                         return false;
1338                 }
1339                 $feed = $ret['body'];
1340                 $dummy1 = null;
1341                 $dummy2 = null;
1342                 $dummy2 = null;
1343                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1344                 if (!$feed_data) {
1345                         return false;
1346                 }
1347
1348                 if (!empty($feed_data["header"]["author-name"])) {
1349                         $data["name"] = $feed_data["header"]["author-name"];
1350                 }
1351                 if (!empty($feed_data["header"]["author-nick"])) {
1352                         $data["nick"] = $feed_data["header"]["author-nick"];
1353                 }
1354                 if (!empty($feed_data["header"]["author-avatar"])) {
1355                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1356                 }
1357                 if (!empty($feed_data["header"]["author-id"])) {
1358                         $data["alias"] = $feed_data["header"]["author-id"];
1359                 }
1360                 if (!empty($feed_data["header"]["author-location"])) {
1361                         $data["location"] = $feed_data["header"]["author-location"];
1362                 }
1363                 if (!empty($feed_data["header"]["author-about"])) {
1364                         $data["about"] = $feed_data["header"]["author-about"];
1365                 }
1366                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1367                 // So we take the value that we just fetched, although the other one worked as well
1368                 if (!empty($feed_data["header"]["author-link"])) {
1369                         $data["url"] = $feed_data["header"]["author-link"];
1370                 }
1371
1372                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1373                         $data['url'] = $data["alias"];
1374                         $data["alias"] = '';
1375                 }
1376
1377                 /// @todo Fetch location and "about" from the feed as well
1378                 return $data;
1379         }
1380
1381         /**
1382          * @brief Fetch data from a pump.io profile page
1383          *
1384          * @param string $profile_link Link to the profile page
1385          *
1386          * @return array profile data
1387          */
1388         private static function pumpioProfileData($profile_link)
1389         {
1390                 $doc = new DOMDocument();
1391                 if (!@$doc->loadHTMLFile($profile_link)) {
1392                         return false;
1393                 }
1394
1395                 $xpath = new DomXPath($doc);
1396
1397                 $data = [];
1398
1399                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1400
1401                 if ($data["name"] == '') {
1402                         // This is ugly - but pump.io doesn't seem to know a better way for it
1403                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1404                         $pos = strpos($data["name"], chr(10));
1405                         if ($pos) {
1406                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1407                         }
1408                 }
1409
1410                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1411
1412                 if ($data["location"] == '') {
1413                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1414                 }
1415
1416                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1417
1418                 if ($data["about"] == '') {
1419                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1420                 }
1421
1422                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1423                 if (!$avatar) {
1424                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1425                 }
1426                 if ($avatar) {
1427                         foreach ($avatar->attributes as $attribute) {
1428                                 if ($attribute->name == "src") {
1429                                         $data["photo"] = trim($attribute->value);
1430                                 }
1431                         }
1432                 }
1433
1434                 return $data;
1435         }
1436
1437         /**
1438          * @brief Check for pump.io contact
1439          *
1440          * @param array $webfinger Webfinger data
1441          *
1442          * @return array pump.io data
1443          */
1444         private static function pumpio($webfinger, $addr)
1445         {
1446                 $data = [];
1447                 foreach ($webfinger["links"] as $link) {
1448                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1449                                 && ($link["type"] == "text/html")
1450                                 && ($link["href"] != "")
1451                         ) {
1452                                 $data["url"] = $link["href"];
1453                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1454                                 $data["notify"] = $link["href"];
1455                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1456                                 $data["poll"] = $link["href"];
1457                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1458                                 $data["dialback"] = $link["href"];
1459                         }
1460                 }
1461                 if (isset($data["poll"]) && isset($data["notify"])
1462                         && isset($data["dialback"])
1463                         && isset($data["url"])
1464                 ) {
1465                         // by now we use these fields only for the network type detection
1466                         // So we unset all data that isn't used at the moment
1467                         unset($data["dialback"]);
1468
1469                         $data["network"] = Protocol::PUMPIO;
1470                 } else {
1471                         return false;
1472                 }
1473
1474                 $profile_data = self::pumpioProfileData($data["url"]);
1475
1476                 if (!$profile_data) {
1477                         return false;
1478                 }
1479
1480                 $data = array_merge($data, $profile_data);
1481
1482                 if (($addr != '') && ($data['name'] != '')) {
1483                         $name = trim(str_replace($addr, '', $data['name']));
1484                         if ($name != '') {
1485                                 $data['name'] = $name;
1486                         }
1487                 }
1488
1489                 return $data;
1490         }
1491
1492         /**
1493          * @brief Check page for feed link
1494          *
1495          * @param string $url Page link
1496          *
1497          * @return string feed link
1498          */
1499         private static function getFeedLink($url)
1500         {
1501                 $doc = new DOMDocument();
1502
1503                 if (!@$doc->loadHTMLFile($url)) {
1504                         return false;
1505                 }
1506
1507                 $xpath = new DomXPath($doc);
1508
1509                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1510                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1511                 if (!is_object($feeds)) {
1512                         return false;
1513                 }
1514
1515                 if ($feeds->length == 0) {
1516                         return false;
1517                 }
1518
1519                 $feed_url = "";
1520
1521                 foreach ($feeds as $feed) {
1522                         $attr = [];
1523                         foreach ($feed->attributes as $attribute) {
1524                                 $attr[$attribute->name] = trim($attribute->value);
1525                         }
1526
1527                         if ($feed_url == "") {
1528                                 $feed_url = $attr["href"];
1529                         }
1530                 }
1531
1532                 return $feed_url;
1533         }
1534
1535         /**
1536          * @brief Check for feed contact
1537          *
1538          * @param string  $url   Profile link
1539          * @param boolean $probe Do a probe if the page contains a feed link
1540          *
1541          * @return array feed data
1542          */
1543         private static function feed($url, $probe = true)
1544         {
1545                 $ret = Network::curl($url);
1546                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1547                         return false;
1548                 }
1549                 $feed = $ret['body'];
1550                 $dummy1 = $dummy2 = $dummy3 = null;
1551                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1552
1553                 if (!$feed_data) {
1554                         if (!$probe) {
1555                                 return false;
1556                         }
1557
1558                         $feed_url = self::getFeedLink($url);
1559
1560                         if (!$feed_url) {
1561                                 return false;
1562                         }
1563
1564                         return self::feed($feed_url, false);
1565                 }
1566
1567                 if (!empty($feed_data["header"]["author-name"])) {
1568                         $data["name"] = $feed_data["header"]["author-name"];
1569                 }
1570
1571                 if (!empty($feed_data["header"]["author-nick"])) {
1572                         $data["nick"] = $feed_data["header"]["author-nick"];
1573                 }
1574
1575                 if (!empty($feed_data["header"]["author-avatar"])) {
1576                         $data["photo"] = $feed_data["header"]["author-avatar"];
1577                 }
1578
1579                 if (!empty($feed_data["header"]["author-id"])) {
1580                         $data["alias"] = $feed_data["header"]["author-id"];
1581                 }
1582
1583                 $data["url"] = $url;
1584                 $data["poll"] = $url;
1585
1586                 if (!empty($feed_data["header"]["author-link"])) {
1587                         $data["baseurl"] = $feed_data["header"]["author-link"];
1588                 } else {
1589                         $data["baseurl"] = $data["url"];
1590                 }
1591
1592                 $data["network"] = Protocol::FEED;
1593
1594                 return $data;
1595         }
1596
1597         /**
1598          * @brief Check for mail contact
1599          *
1600          * @param string  $uri Profile link
1601          * @param integer $uid User ID
1602          *
1603          * @return array mail data
1604          */
1605         private static function mail($uri, $uid)
1606         {
1607                 if (!Network::isEmailDomainValid($uri)) {
1608                         return false;
1609                 }
1610
1611                 if ($uid == 0) {
1612                         return false;
1613                 }
1614
1615                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1616
1617                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1618                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1619                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1620
1621                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1622                         return false;
1623                 }
1624
1625                 $mailbox = Email::constructMailboxName($mailacct);
1626                 $password = '';
1627                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1628                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1629                 if (!$mbox) {
1630                         return false;
1631                 }
1632
1633                 $msgs = Email::poll($mbox, $uri);
1634                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1635
1636                 if (!count($msgs)) {
1637                         return false;
1638                 }
1639
1640                 $phost = substr($uri, strpos($uri, '@') + 1);
1641
1642                 $data = [];
1643                 $data["addr"]    = $uri;
1644                 $data["network"] = Protocol::MAIL;
1645                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1646                 $data["nick"]    = $data["name"];
1647                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1648                 $data["url"]     = 'mailto:'.$uri;
1649                 $data["notify"]  = 'smtp '.random_string();
1650                 $data["poll"]    = 'email '.random_string();
1651
1652                 $x = Email::messageMeta($mbox, $msgs[0]);
1653                 if (stristr($x[0]->from, $uri)) {
1654                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1655                 } elseif (stristr($x[0]->to, $uri)) {
1656                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1657                 }
1658                 if (isset($adr)) {
1659                         foreach ($adr as $feadr) {
1660                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1661                                         &&(strcasecmp($feadr->host, $phost) == 0)
1662                                         && (strlen($feadr->personal))
1663                                 ) {
1664                                         $personal = imap_mime_header_decode($feadr->personal);
1665                                         $data["name"] = "";
1666                                         foreach ($personal as $perspart) {
1667                                                 if ($perspart->charset != "default") {
1668                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1669                                                 } else {
1670                                                         $data["name"] .= $perspart->text;
1671                                                 }
1672                                         }
1673
1674                                         $data["name"] = notags($data["name"]);
1675                                 }
1676                         }
1677                 }
1678                 if (!empty($mbox)) {
1679                         imap_close($mbox);
1680                 }
1681                 return $data;
1682         }
1683
1684         /**
1685          * @brief Mix two paths together to possibly fix missing parts
1686          *
1687          * @param string $avatar Path to the avatar
1688          * @param string $base   Another path that is hopefully complete
1689          *
1690          * @return string fixed avatar path
1691          */
1692         public static function fixAvatar($avatar, $base)
1693         {
1694                 $base_parts = parse_url($base);
1695
1696                 // Remove all parts that could create a problem
1697                 unset($base_parts['path']);
1698                 unset($base_parts['query']);
1699                 unset($base_parts['fragment']);
1700
1701                 $avatar_parts = parse_url($avatar);
1702
1703                 // Now we mix them
1704                 $parts = array_merge($base_parts, $avatar_parts);
1705
1706                 // And put them together again
1707                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1708                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1709                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1710                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1711                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1712                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1713
1714                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1715
1716                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1717
1718                 return $fixed;
1719         }
1720 }