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