src/Network/Probe.php

   1 <?php
   2
   3 namespace Friendica\Network;
   4
   5 /**
   6  * @file src/Network/Probe.php
   7  * @brief Functions for probing URL
   8  *
   9  */
  10
  11 use Friendica\App;
  12 use Friendica\Core\System;
  13 use Friendica\Core\Config;
  14
  15 use dbm;
  16 use Cache;
  17 use xml;
  18
  19 use DomXPath;
  20 use DOMDocument;
  21
  22 require_once 'include/feed.php';
  23 require_once 'include/email.php';
  24 require_once 'include/network.php';
  25
  26 /**
  27  * @brief This class contain functions for probing URL
  28  *
  29  */
  30 class Probe {
  31
  32         private static $baseurl;
  33
  34         /**
  35          * @brief Rearrange the array so that it always has the same order
  36          *
  37          * @param array $data Unordered data
  38          *
  39          * @return array Ordered data
  40          */
  41         private static function rearrangeData($data) {
  42                 $fields = array("name", "nick", "guid", "url", "addr", "alias",
  43                                 "photo", "community", "keywords", "location", "about",
  44                                 "batch", "notify", "poll", "request", "confirm", "poco",
  45                                 "priority", "network", "pubkey", "baseurl");
  46
  47                 $newdata = array();
  48                 foreach ($fields as $field) {
  49                         if (isset($data[$field])) {
  50                                 $newdata[$field] = $data[$field];
  51                         } else {
  52                                 $newdata[$field] = "";
  53                         }
  54                 }
  55
  56                 // We don't use the "priority" field anymore and replace it with a dummy.
  57                 $newdata["priority"] = 0;
  58
  59                 return $newdata;
  60         }
  61
  62         /**
  63          * @brief Check if the hostname belongs to the own server
  64          *
  65          * @param string $host The hostname that is to be checked
  66          *
  67          * @return bool Does the testes hostname belongs to the own server?
  68          */
  69         private static function ownHost($host) {
  70                 $own_host = get_app()->get_hostname();
  71
  72                 $parts = parse_url($host);
  73
  74                 if (!isset($parts['scheme'])) {
  75                         $parts = parse_url('http://'.$host);
  76                 }
  77
  78                 if (!isset($parts['host'])) {
  79                         return false;
  80                 }
  81                 return $parts['host'] == $own_host;
  82         }
  83
  84         /**
  85          * @brief Probes for XRD data
  86          *
  87          * @param string $host The host part of an url
  88          *
  89          * @return array
  90          *      'lrdd' => Link to LRDD endpoint
  91          *      'lrdd-xml' => Link to LRDD endpoint in XML format
  92          *      'lrdd-json' => Link to LRDD endpoint in JSON format
  93          */
  94         private static function xrd($host) {
  95
  96                 // Reset the static variable
  97                 self::$baseurl = '';
  98
  99                 $ssl_url = "https://".$host."/.well-known/host-meta";
 100                 $url = "http://".$host."/.well-known/host-meta";
 101
 102                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 103                 $redirects = 0;
 104
 105                 logger("Probing for ".$host, LOGGER_DEBUG);
 106
 107                 $ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 108                 if ($ret['success']) {
 109                         $xml = $ret['body'];
 110                         $xrd = parse_xml_string($xml, false);
 111                 }
 112
 113                 if (!is_object($xrd)) {
 114                         $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 115                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 116                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 117                                 return false;
 118                         }
 119                         $xml = $ret['body'];
 120                         $xrd = parse_xml_string($xml, false);
 121                 }
 122                 if (!is_object($xrd)) {
 123                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 124                         return array();
 125                 }
 126
 127                 $links = xml::element_to_array($xrd);
 128                 if (!isset($links["xrd"]["link"])) {
 129                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 130                         return array();
 131                 }
 132
 133                 $xrd_data = array();
 134
 135                 foreach ($links["xrd"]["link"] as $value => $link) {
 136                         if (!empty($link["@attributes"])) {
 137                                 $attributes = $link["@attributes"];
 138                         } elseif ($value == "@attributes") {
 139                                 $attributes = $link;
 140                         } else {
 141                                 continue;
 142                         }
 143
 144                         if (($attributes["rel"] == "lrdd")
 145                                 && ($attributes["type"] == "application/xrd+xml")
 146                         ) {
 147                                 $xrd_data["lrdd-xml"] = $attributes["template"];
 148                         } elseif (($attributes["rel"] == "lrdd")
 149                                 && ($attributes["type"] == "application/json")
 150                         ) {
 151                                 $xrd_data["lrdd-json"] = $attributes["template"];
 152                         } elseif ($attributes["rel"] == "lrdd") {
 153                                 $xrd_data["lrdd"] = $attributes["template"];
 154                         }
 155                 }
 156
 157                 self::$baseurl = "http://".$host;
 158
 159                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 160
 161                 return $xrd_data;
 162         }
 163
 164         /**
 165          * @brief Perform Webfinger lookup and return DFRN data
 166          *
 167          * Given an email style address, perform webfinger lookup and
 168          * return the resulting DFRN profile URL, or if no DFRN profile URL
 169          * is located, returns an OStatus subscription template (prefixed
 170          * with the string 'stat:' to identify it as on OStatus template).
 171          * If this isn't an email style address just return $webbie.
 172          * Return an empty string if email-style addresses but webfinger fails,
 173          * or if the resultant personal XRD doesn't contain a supported
 174          * subscription/friend-request attribute.
 175          *
 176          * amended 7/9/2011 to return an hcard which could save potentially loading
 177          * a lengthy content page to scrape dfrn attributes
 178          *
 179          * @param string $webbie Address that should be probed
 180          * @param string $hcard_url Link to the hcard - is returned by reference
 181          *
 182          * @return string profile link
 183          */
 184         public static function webfingerDfrn($webbie, &$hcard_url) {
 185
 186                 $profile_link = '';
 187
 188                 $links = self::lrdd($webbie);
 189                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 190                 if (count($links)) {
 191                         foreach ($links as $link) {
 192                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 193                                         $profile_link = $link['@attributes']['href'];
 194                                 }
 195                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 196                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 197                                 }
 198                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 199                                         $hcard_url = $link['@attributes']['href'];
 200                                 }
 201                         }
 202                 }
 203                 return $profile_link;
 204         }
 205
 206         /**
 207          * @brief Check an URI for LRDD data
 208          *
 209          * this is a replacement for the "lrdd" function in include/network.php.
 210          * It isn't used in this class and has some redundancies in the code.
 211          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 212          *
 213          * @param string $uri Address that should be probed
 214          *
 215          * @return array uri data
 216          */
 217         public static function lrdd($uri) {
 218
 219                 $lrdd = self::xrd($uri);
 220                 $webfinger = null;
 221
 222                 if (is_bool($lrdd)) {
 223                         return array();
 224                 }
 225
 226                 if (!$lrdd) {
 227                         $parts = @parse_url($uri);
 228                         if (!$parts) {
 229                                 return array();
 230                         }
 231
 232                         $host = $parts["host"];
 233                         if (!empty($parts["port"])) {
 234                                 $host .= ':'.$parts["port"];
 235                         }
 236
 237                         $path_parts = explode("/", trim($parts["path"], "/"));
 238
 239                         $nick = array_pop($path_parts);
 240
 241                         do {
 242                                 $lrdd = self::xrd($host);
 243                                 $host .= "/".array_shift($path_parts);
 244                         } while (!$lrdd && (sizeof($path_parts) > 0));
 245                 }
 246
 247                 if (!$lrdd) {
 248                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 249                         return array();
 250                 }
 251
 252                 foreach ($lrdd as $key => $link) {
 253                         if ($webfinger) {
 254                                 continue;
 255                         }
 256
 257                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
 258                                 continue;
 259                         }
 260
 261                         $path = str_replace('{uri}', urlencode($uri), $link);
 262                         $webfinger = self::webfinger($path);
 263
 264                         if (!$webfinger && (strstr($uri, "@"))) {
 265                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
 266                                 $webfinger = self::webfinger($path);
 267                         }
 268
 269                         // Special treatment for Mastodon
 270                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 271                         // But the webfinger for this format fails.
 272                         if (!$webfinger && !empty($nick)) {
 273                                 // Mastodon uses a "@" as prefix for usernames in their url format
 274                                 $nick = ltrim($nick, '@');
 275
 276                                 $addr = $nick."@".$host;
 277
 278                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 279                                 $webfinger = self::webfinger($path);
 280                         }
 281                 }
 282
 283                 if (!is_array($webfinger["links"])) {
 284                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 285                         return false;
 286                 }
 287
 288                 $data = array();
 289
 290                 foreach ($webfinger["links"] as $link) {
 291                         $data[] = array("@attributes" => $link);
 292                 }
 293
 294                 if (is_array($webfinger["aliases"])) {
 295                         foreach ($webfinger["aliases"] as $alias) {
 296                                 $data[] = array("@attributes" =>
 297                                                         array("rel" => "alias",
 298                                                                 "href" => $alias));
 299                         }
 300                 }
 301
 302                 return $data;
 303         }
 304
 305         /**
 306          * @brief Fetch information (protocol endpoints and user information) about a given uri
 307          *
 308          * @param string $uri Address that should be probed
 309          * @param string $network Test for this specific network
 310          * @param integer $uid User ID for the probe (only used for mails)
 311          * @param boolean $cache Use cached values?
 312          *
 313          * @return array uri data
 314          */
 315         public static function uri($uri, $network = "", $uid = 0, $cache = true) {
 316
 317                 if ($cache) {
 318                         $result = Cache::get("probe_url:".$network.":".$uri);
 319                         if (!is_null($result)) {
 320                                 return $result;
 321                         }
 322                 }
 323
 324                 if ($uid == 0) {
 325                         $uid = local_user();
 326                 }
 327
 328                 $data = self::detect($uri, $network, $uid);
 329
 330                 if (!isset($data["url"])) {
 331                         $data["url"] = $uri;
 332                 }
 333
 334                 if ($data["photo"] != "") {
 335                         $data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
 336                 } else {
 337                         $data["photo"] = App::get_baseurl().'/images/person-175.jpg';
 338                 }
 339
 340                 if (empty($data["name"])) {
 341                         if (!empty($data["nick"])) {
 342                                 $data["name"] = $data["nick"];
 343                         }
 344
 345                         if ($data["name"] == "") {
 346                                 $data["name"] = $data["url"];
 347                         }
 348                 }
 349
 350                 if (empty($data["nick"])) {
 351                         $data["nick"] = strtolower($data["name"]);
 352
 353                         if (strpos($data['nick'], ' ')) {
 354                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 355                         }
 356                 }
 357
 358                 if (self::$baseurl != "") {
 359                         $data["baseurl"] = self::$baseurl;
 360                 }
 361
 362                 if (!isset($data["network"])) {
 363                         $data["network"] = NETWORK_PHANTOM;
 364                 }
 365
 366                 $data = self::rearrangeData($data);
 367
 368                 // Only store into the cache if the value seems to be valid
 369                 if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
 370                         Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
 371
 372                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 373                         /// The biggest problem is the avatar picture that could have a reduced image size.
 374                         /// It should only be updated if the existing picture isn't existing anymore.
 375                         /// We only update the contact when it is no probing for a specific network.
 376                         if (($data['network'] != NETWORK_FEED)
 377                                 && ($network == "")
 378                                 && $data["name"]
 379                                 && $data["nick"]
 380                                 && $data["url"]
 381                                 && $data["addr"]
 382                                 && $data["poll"]
 383                         ) {
 384                                 q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 385                                                 `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 386                                         WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 387                                         dbesc($data["name"]),
 388                                         dbesc($data["nick"]),
 389                                         dbesc($data["url"]),
 390                                         dbesc($data["addr"]),
 391                                         dbesc($data["notify"]),
 392                                         dbesc($data["poll"]),
 393                                         dbesc($data["alias"]),
 394                                         dbesc(datetime_convert()),
 395                                         dbesc(normalise_link($data['url']))
 396                                 );
 397                         }
 398                 }
 399
 400                 return $data;
 401         }
 402
 403         /**
 404          * @brief Switch the scheme of an url between http and https
 405          *
 406          * @param string $url URL
 407          *
 408          * @return string switched URL
 409          */
 410         private static function switchScheme($url) {
 411                 $parts = parse_url($url);
 412
 413                 if (!isset($parts['scheme'])) {
 414                         return $url;
 415                 }
 416
 417                 if ($parts['scheme'] == 'http') {
 418                         $url = str_replace('http://', 'https://', $url);
 419                 } elseif ($parts['scheme'] == 'https') {
 420                         $url = str_replace('https://', 'http://', $url);
 421                 }
 422
 423                 return $url;
 424         }
 425
 426         /**
 427          * @brief Checks if a profile url should be OStatus but only provides partial information
 428          *
 429          * @param array $webfinger Webfinger data
 430          * @param string $lrdd Path template for webfinger request
 431          *
 432          * @return array fixed webfinger data
 433          */
 434         private static function fixOstatus($webfinger, $lrdd) {
 435                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 436                         return $webfinger;
 437                 }
 438
 439                 $is_ostatus = false;
 440                 $has_key = false;
 441
 442                 foreach ($webfinger['links'] as $link) {
 443                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 444                                 $is_ostatus = true;
 445                         }
 446                         if ($link['rel'] == 'magic-public-key') {
 447                                 $has_key = true;
 448                         }
 449                 }
 450
 451                 if (!$is_ostatus || $has_key) {
 452                         return $webfinger;
 453                 }
 454
 455                 $url = self::switchScheme($webfinger['subject']);
 456                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 457                 $webfinger2 = self::webfinger($path);
 458
 459                 // Is the new webfinger detectable as OStatus?
 460                 if (self::ostatus($webfinger2, true)) {
 461                         $webfinger = $webfinger2;
 462                 }
 463
 464                 return $webfinger;
 465         }
 466
 467         /**
 468          * @brief Fetch information (protocol endpoints and user information) about a given uri
 469          *
 470          * This function is only called by the "uri" function that adds caching and rearranging of data.
 471          *
 472          * @param string $uri Address that should be probed
 473          * @param string $network Test for this specific network
 474          * @param integer $uid User ID for the probe (only used for mails)
 475          *
 476          * @return array uri data
 477          */
 478         private static function detect($uri, $network, $uid) {
 479                 $parts = parse_url($uri);
 480
 481                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 482                         $host = $parts["host"];
 483                         if (!empty($parts["port"])) {
 484                                 $host .= ':'.$parts["port"];
 485                         }
 486
 487                         if ($host == 'twitter.com') {
 488                                 return array("network" => NETWORK_TWITTER);
 489                         }
 490                         $lrdd = self::xrd($host);
 491
 492                         if (is_bool($lrdd)) {
 493                                 return array();
 494                         }
 495
 496                         $path_parts = explode("/", trim($parts["path"], "/"));
 497
 498                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 499                                 $host .= "/".array_shift($path_parts);
 500                                 $lrdd = self::xrd($host);
 501                         }
 502                         if (!$lrdd) {
 503                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 504                                 return self::feed($uri);
 505                         }
 506                         $nick = array_pop($path_parts);
 507
 508                         // Mastodon uses a "@" as prefix for usernames in their url format
 509                         $nick = ltrim($nick, '@');
 510
 511                         $addr = $nick."@".$host;
 512
 513                 } elseif (strstr($uri, '@')) {
 514                         // If the URI starts with "mailto:" then jump directly to the mail detection
 515                         if (strpos($uri, 'mailto:') !== false) {
 516                                 $uri = str_replace('mailto:', '', $uri);
 517                                 return self::mail($uri, $uid);
 518                         }
 519
 520                         if ($network == NETWORK_MAIL) {
 521                                 return self::mail($uri, $uid);
 522                         }
 523                         // Remove "acct:" from the URI
 524                         $uri = str_replace('acct:', '', $uri);
 525
 526                         $host = substr($uri, strpos($uri, '@') + 1);
 527                         $nick = substr($uri, 0, strpos($uri, '@'));
 528
 529                         if (strpos($uri, '@twitter.com')) {
 530                                 return array("network" => NETWORK_TWITTER);
 531                         }
 532                         $lrdd = self::xrd($host);
 533
 534                         if (is_bool($lrdd)) {
 535                                 return array();
 536                         }
 537
 538                         if (!$lrdd) {
 539                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 540                                 return self::mail($uri, $uid);
 541                         }
 542                         $addr = $uri;
 543
 544                 } else {
 545                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 546                         return false;
 547                 }
 548
 549                 $webfinger = false;
 550
 551                 /// @todo Do we need the prefix "acct:" or "acct://"?
 552
 553                 foreach ($lrdd as $key => $link) {
 554                         if ($webfinger) {
 555                                 continue;
 556                         }
 557                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
 558                                 continue;
 559                         }
 560                         // At first try it with the given uri
 561                         $path = str_replace('{uri}', urlencode($uri), $link);
 562                         $webfinger = self::webfinger($path);
 563
 564                         // Fix possible problems with GNU Social probing to wrong scheme
 565                         $webfinger = self::fixOstatus($webfinger, $link);
 566
 567                         // We cannot be sure that the detected address was correct, so we don't use the values
 568                         if ($webfinger && ($uri != $addr)) {
 569                                 $nick = "";
 570                                 $addr = "";
 571                         }
 572
 573                         // Try webfinger with the address (user@domain.tld)
 574                         if (!$webfinger) {
 575                                 $path = str_replace('{uri}', urlencode($addr), $link);
 576                                 $webfinger = self::webfinger($path);
 577                         }
 578
 579                         // Mastodon needs to have it with "acct:"
 580                         if (!$webfinger) {
 581                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 582                                 $webfinger = self::webfinger($path);
 583                         }
 584                 }
 585                 if (!$webfinger) {
 586                         return self::feed($uri);
 587                 }
 588
 589                 $result = false;
 590
 591                 logger("Probing ".$uri, LOGGER_DEBUG);
 592
 593                 if (in_array($network, array("", NETWORK_DFRN))) {
 594                         $result = self::dfrn($webfinger);
 595                 }
 596                 if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
 597                         $result = self::diaspora($webfinger);
 598                 }
 599                 if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
 600                         $result = self::ostatus($webfinger);
 601                 }
 602                 if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
 603                         $result = self::pumpio($webfinger);
 604                 }
 605                 if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
 606                         $result = self::feed($uri);
 607                 } else {
 608                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 609                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 610                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 611                                 $result["nick"] = $nick;
 612                         }
 613
 614                         if (empty($result["addr"]) && ($addr != "")) {
 615                                 $result["addr"] = $addr;
 616                         }
 617                 }
 618
 619                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 620
 621                 if (empty($result["baseurl"])) {
 622                         $pos = strpos($result["url"], $host);
 623                         if ($pos) {
 624                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 625                         }
 626                 }
 627
 628                 return $result;
 629         }
 630
 631         /**
 632          * @brief Perform a webfinger request.
 633          *
 634          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 635          *
 636          * @param string $url Address that should be probed
 637          *
 638          * @return array webfinger data
 639          */
 640         private static function webfinger($url) {
 641
 642                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 643                 $redirects = 0;
 644
 645                 $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 646                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 647                         return false;
 648                 }
 649                 $data = $ret['body'];
 650
 651                 $xrd = parse_xml_string($data, false);
 652
 653                 if (!is_object($xrd)) {
 654                         // If it is not XML, maybe it is JSON
 655                         $webfinger = json_decode($data, true);
 656
 657                         if (!isset($webfinger["links"])) {
 658                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 659                                 return false;
 660                         }
 661
 662                         return $webfinger;
 663                 }
 664
 665                 $xrd_arr = xml::element_to_array($xrd);
 666                 if (!isset($xrd_arr["xrd"]["link"])) {
 667                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 668                         return false;
 669                 }
 670
 671                 $webfinger = array();
 672
 673                 if (!empty($xrd_arr["xrd"]["subject"])) {
 674                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 675                 }
 676
 677                 if (!empty($xrd_arr["xrd"]["alias"])) {
 678                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 679                 }
 680
 681                 $webfinger["links"] = array();
 682
 683                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 684                         if (!empty($data["@attributes"])) {
 685                                 $attributes = $data["@attributes"];
 686                         } elseif ($value == "@attributes") {
 687                                 $attributes = $data;
 688                         } else {
 689                                 continue;
 690                         }
 691
 692                         $webfinger["links"][] = $attributes;
 693                 }
 694                 return $webfinger;
 695         }
 696
 697         /**
 698          * @brief Poll the Friendica specific noscrape page.
 699          *
 700          * "noscrape" is a faster alternative to fetch the data from the hcard.
 701          * This functionality was originally created for the directory.
 702          *
 703          * @param string $noscrape_url Link to the noscrape page
 704          * @param array $data The already fetched data
 705          *
 706          * @return array noscrape data
 707          */
 708         private static function pollNoscrape($noscrape_url, $data) {
 709                 $ret = z_fetch_url($noscrape_url);
 710                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 711                         return false;
 712                 }
 713                 $content = $ret['body'];
 714                 if (!$content) {
 715                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 716                         return false;
 717                 }
 718
 719                 $json = json_decode($content, true);
 720                 if (!is_array($json)) {
 721                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 722                         return false;
 723                 }
 724
 725                 if (!empty($json["fn"])) {
 726                         $data["name"] = $json["fn"];
 727                 }
 728
 729                 if (!empty($json["addr"])) {
 730                         $data["addr"] = $json["addr"];
 731                 }
 732
 733                 if (!empty($json["nick"])) {
 734                         $data["nick"] = $json["nick"];
 735                 }
 736
 737                 if (!empty($json["comm"])) {
 738                         $data["community"] = $json["comm"];
 739                 }
 740
 741                 if (!empty($json["tags"])) {
 742                         $keywords = implode(" ", $json["tags"]);
 743                         if ($keywords != "") {
 744                                 $data["keywords"] = $keywords;
 745                         }
 746                 }
 747
 748                 $location = formatted_location($json);
 749                 if ($location) {
 750                         $data["location"] = $location;
 751                 }
 752
 753                 if (!empty($json["about"])) {
 754                         $data["about"] = $json["about"];
 755                 }
 756
 757                 if (!empty($json["key"])) {
 758                         $data["pubkey"] = $json["key"];
 759                 }
 760
 761                 if (!empty($json["photo"])) {
 762                         $data["photo"] = $json["photo"];
 763                 }
 764
 765                 if (!empty($json["dfrn-request"])) {
 766                         $data["request"] = $json["dfrn-request"];
 767                 }
 768
 769                 if (!empty($json["dfrn-confirm"])) {
 770                         $data["confirm"] = $json["dfrn-confirm"];
 771                 }
 772
 773                 if (!empty($json["dfrn-notify"])) {
 774                         $data["notify"] = $json["dfrn-notify"];
 775                 }
 776
 777                 if (!empty($json["dfrn-poll"])) {
 778                         $data["poll"] = $json["dfrn-poll"];
 779                 }
 780
 781                 return $data;
 782         }
 783
 784         /**
 785          * @brief Check for valid DFRN data
 786          *
 787          * @param array $data DFRN data
 788          *
 789          * @return int Number of errors
 790          */
 791         public static function validDfrn($data) {
 792                 $errors = 0;
 793                 if (!isset($data['key'])) {
 794                         $errors ++;
 795                 }
 796                 if (!isset($data['dfrn-request'])) {
 797                         $errors ++;
 798                 }
 799                 if (!isset($data['dfrn-confirm'])) {
 800                         $errors ++;
 801                 }
 802                 if (!isset($data['dfrn-notify'])) {
 803                         $errors ++;
 804                 }
 805                 if (!isset($data['dfrn-poll'])) {
 806                         $errors ++;
 807                 }
 808                 return $errors;
 809         }
 810
 811         /**
 812          * @brief Fetch data from a DFRN profile page and via "noscrape"
 813          *
 814          * @param string $profile_link Link to the profile page
 815          *
 816          * @return array profile data
 817          */
 818         public static function profile($profile_link) {
 819
 820                 $data = array();
 821
 822                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 823
 824                 // Fetch data via noscrape - this is faster
 825                 $noscrape_url = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile_link);
 826                 $data = self::pollNoscrape($noscrape_url, $data);
 827
 828                 if (!isset($data["notify"])
 829                         || !isset($data["confirm"])
 830                         || !isset($data["request"])
 831                         || !isset($data["poll"])
 832                         || !isset($data["poco"])
 833                         || !isset($data["name"])
 834                         || !isset($data["photo"])
 835                 ) {
 836                         $data = self::pollHcard($profile_link, $data, true);
 837                 }
 838
 839                 $prof_data = array();
 840                 $prof_data["addr"]         = $data["addr"];
 841                 $prof_data["nick"]         = $data["nick"];
 842                 $prof_data["dfrn-request"] = $data["request"];
 843                 $prof_data["dfrn-confirm"] = $data["confirm"];
 844                 $prof_data["dfrn-notify"]  = $data["notify"];
 845                 $prof_data["dfrn-poll"]    = $data["poll"];
 846                 $prof_data["dfrn-poco"]    = $data["poco"];
 847                 $prof_data["photo"]        = $data["photo"];
 848                 $prof_data["fn"]           = $data["name"];
 849                 $prof_data["key"]          = $data["pubkey"];
 850
 851                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 852
 853                 return $prof_data;
 854         }
 855
 856         /**
 857          * @brief Check for DFRN contact
 858          *
 859          * @param array $webfinger Webfinger data
 860          *
 861          * @return array DFRN data
 862          */
 863         private static function dfrn($webfinger) {
 864
 865                 $hcard_url = "";
 866                 $data = array();
 867                 foreach ($webfinger["links"] as $link) {
 868                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 869                                 $data["network"] = NETWORK_DFRN;
 870                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 871                                 $data["poll"] = $link["href"];
 872                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 873                                 $data["url"] = $link["href"];
 874                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 875                                 $hcard_url = $link["href"];
 876                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 877                                 $data["poco"] = $link["href"];
 878                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 879                                 $data["photo"] = $link["href"];
 880                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 881                                 $data["baseurl"] = trim($link["href"], '/');
 882                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 883                                 $data["guid"] = $link["href"];
 884                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 885                                 $data["pubkey"] = base64_decode($link["href"]);
 886
 887                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 888                                 if (strstr($data["pubkey"], 'RSA ')) {
 889                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 890                                 }
 891                         }
 892                 }
 893
 894                 if (is_array($webfinger["aliases"])) {
 895                         foreach ($webfinger["aliases"] as $alias) {
 896                                 if (substr($alias, 0, 5) == 'acct:') {
 897                                         $data["addr"] = substr($alias, 5);
 898                                 }
 899                         }
 900                 }
 901
 902                 if (!isset($data["network"]) || ($hcard_url == "")) {
 903                         return false;
 904                 }
 905
 906                 // Fetch data via noscrape - this is faster
 907                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
 908                 $data = self::pollNoscrape($noscrape_url, $data);
 909
 910                 if (isset($data["notify"])
 911                         && isset($data["confirm"])
 912                         && isset($data["request"])
 913                         && isset($data["poll"])
 914                         && isset($data["name"])
 915                         && isset($data["photo"])
 916                 ) {
 917                         return $data;
 918                 }
 919
 920                 $data = self::pollHcard($hcard_url, $data, true);
 921
 922                 return $data;
 923         }
 924
 925         /**
 926          * @brief Poll the hcard page (Diaspora and Friendica specific)
 927          *
 928          * @param string $hcard_url Link to the hcard page
 929          * @param array $data The already fetched data
 930          * @param boolean $dfrn Poll DFRN specific data
 931          *
 932          * @return array hcard data
 933          */
 934         private static function pollHcard($hcard_url, $data, $dfrn = false) {
 935                 $ret = z_fetch_url($hcard_url);
 936                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 937                         return false;
 938                 }
 939                 $content = $ret['body'];
 940                 if (!$content) {
 941                         return false;
 942                 }
 943
 944                 $doc = new DOMDocument();
 945                 if (!@$doc->loadHTML($content)) {
 946                         return false;
 947                 }
 948
 949                 $xpath = new DomXPath($doc);
 950
 951                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 952                 if (!is_object($vcards)) {
 953                         return false;
 954                 }
 955
 956                 if ($vcards->length > 0) {
 957                         $vcard = $vcards->item(0);
 958
 959                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 960                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 961                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 962                         if (($search->length > 0) && ($data["guid"] == "")) {
 963                                 $data["guid"] = $search->item(0)->nodeValue;
 964                         }
 965
 966                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 967                         if ($search->length > 0) {
 968                                 $data["nick"] = $search->item(0)->nodeValue;
 969                         }
 970
 971                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 972                         if ($search->length > 0) {
 973                                 $data["name"] = $search->item(0)->nodeValue;
 974                         }
 975
 976                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
 977                         if ($search->length > 0) {
 978                                 $data["searchable"] = $search->item(0)->nodeValue;
 979                         }
 980
 981                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
 982                         if ($search->length > 0) {
 983                                 $data["pubkey"] = $search->item(0)->nodeValue;
 984                                 if (strstr($data["pubkey"], 'RSA ')) {
 985                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 986                                 }
 987                         }
 988
 989                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
 990                         if ($search->length > 0) {
 991                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
 992                         }
 993                 }
 994
 995                 $avatar = array();
 996                 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
 997                 foreach ($photos as $photo) {
 998                         $attr = array();
 999                         foreach ($photo->attributes as $attribute) {
1000                                 $attr[$attribute->name] = trim($attribute->value);
1001                         }
1002
1003                         if (isset($attr["src"]) && isset($attr["width"])) {
1004                                 $avatar[$attr["width"]] = $attr["src"];
1005                         }
1006
1007                         // We don't have a width. So we just take everything that we got.
1008                         // This is a Hubzilla workaround which doesn't send a width.
1009                         if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1010                                 $avatar[] = $attr["src"];
1011                         }
1012                 }
1013
1014                 if (sizeof($avatar)) {
1015                         ksort($avatar);
1016                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1017                 }
1018
1019                 if ($dfrn) {
1020                         // Poll DFRN specific data
1021                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1022                         if ($search->length > 0) {
1023                                 foreach ($search as $link) {
1024                                         //$data["request"] = $search->item(0)->nodeValue;
1025                                         $attr = array();
1026                                         foreach ($link->attributes as $attribute) {
1027                                                 $attr[$attribute->name] = trim($attribute->value);
1028                                         }
1029
1030                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1031                                 }
1032                         }
1033
1034                         // Older Friendica versions had used the "uid" field differently than newer versions
1035                         if ($data["nick"] == $data["guid"]) {
1036                                 unset($data["guid"]);
1037                         }
1038                 }
1039
1040
1041                 return $data;
1042         }
1043
1044         /**
1045          * @brief Check for Diaspora contact
1046          *
1047          * @param array $webfinger Webfinger data
1048          *
1049          * @return array Diaspora data
1050          */
1051         private static function diaspora($webfinger) {
1052
1053                 $hcard_url = "";
1054                 $data = array();
1055                 foreach ($webfinger["links"] as $link) {
1056                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1057                                 $hcard_url = $link["href"];
1058                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1059                                 $data["baseurl"] = trim($link["href"], '/');
1060                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1061                                 $data["guid"] = $link["href"];
1062                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1063                                 $data["url"] = $link["href"];
1064                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1065                                 $data["poll"] = $link["href"];
1066                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1067                                 $data["poco"] = $link["href"];
1068                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1069                                 $data["notify"] = $link["href"];
1070                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1071                                 $data["pubkey"] = base64_decode($link["href"]);
1072
1073                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1074                                 if (strstr($data["pubkey"], 'RSA ')) {
1075                                         $data["pubkey"] = rsatopem($data["pubkey"]);
1076                                 }
1077                         }
1078                 }
1079
1080                 if (!isset($data["url"]) || ($hcard_url == "")) {
1081                         return false;
1082                 }
1083
1084                 if (is_array($webfinger["aliases"])) {
1085                         foreach ($webfinger["aliases"] as $alias) {
1086                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1087                                         $data["alias"] = $alias;
1088                                 }
1089                         }
1090                 }
1091
1092                 // Fetch further information from the hcard
1093                 $data = self::pollHcard($hcard_url, $data);
1094
1095                 if (!$data) {
1096                         return false;
1097                 }
1098
1099                 if (isset($data["url"])
1100                         && isset($data["guid"])
1101                         && isset($data["baseurl"])
1102                         && isset($data["pubkey"])
1103                         && ($hcard_url != "")
1104                 ) {
1105                         $data["network"] = NETWORK_DIASPORA;
1106
1107                         // The Diaspora handle must always be lowercase
1108                         $data["addr"] = strtolower($data["addr"]);
1109
1110                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1111                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1112                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1113                 } else {
1114                         return false;
1115                 }
1116
1117                 return $data;
1118         }
1119
1120         /**
1121          * @brief Check for OStatus contact
1122          *
1123          * @param array $webfinger Webfinger data
1124          * @param bool $short Short detection mode
1125          *
1126          * @return array|bool OStatus data or "false" on error or "true" on short mode
1127          */
1128         private static function ostatus($webfinger, $short = false) {
1129                 $data = array();
1130                 if (is_array($webfinger["aliases"])) {
1131                         foreach ($webfinger["aliases"] as $alias) {
1132                                 if (strstr($alias, "@")) {
1133                                         $data["addr"] = str_replace('acct:', '', $alias);
1134                                 }
1135                         }
1136                 }
1137
1138                 if (is_string($webfinger["subject"]) && strstr($webfinger["subject"], "@")) {
1139                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1140                 }
1141                 $pubkey = "";
1142                 foreach ($webfinger["links"] as $link) {
1143                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1144                                 && ($link["type"] == "text/html")
1145                                 && ($link["href"] != "")
1146                         ) {
1147                                 $data["url"] = $link["href"];
1148                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1149                                 $data["notify"] = $link["href"];
1150                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1151                                 $data["poll"] = $link["href"];
1152                         } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1153                                 $pubkey = $link["href"];
1154
1155                                 if (substr($pubkey, 0, 5) === 'data:') {
1156                                         if (strstr($pubkey, ',')) {
1157                                                 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1158                                         } else {
1159                                                 $pubkey = substr($pubkey, 5);
1160                                         }
1161                                 } elseif (normalise_link($pubkey) == 'http://') {
1162                                         $ret = z_fetch_url($pubkey);
1163                                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1164                                                 return false;
1165                                         }
1166                                         $pubkey = $ret['body'];
1167                                 }
1168
1169                                 $key = explode(".", $pubkey);
1170
1171                                 if (sizeof($key) >= 3) {
1172                                         $m = base64url_decode($key[1]);
1173                                         $e = base64url_decode($key[2]);
1174                                         $data["pubkey"] = metopem($m, $e);
1175                                 }
1176                         }
1177                 }
1178
1179                 if (isset($data["notify"]) && isset($data["pubkey"])
1180                         && isset($data["poll"])
1181                         && isset($data["url"])
1182                 ) {
1183                         $data["network"] = NETWORK_OSTATUS;
1184                 } else {
1185                         return false;
1186                 }
1187
1188                 if ($short) {
1189                         return true;
1190                 }
1191
1192                 // Fetch all additional data from the feed
1193                 $ret = z_fetch_url($data["poll"]);
1194                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1195                         return false;
1196                 }
1197                 $feed = $ret['body'];
1198                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1199                 if (!$feed_data) {
1200                         return false;
1201                 }
1202
1203                 if ($feed_data["header"]["author-name"] != "") {
1204                         $data["name"] = $feed_data["header"]["author-name"];
1205                 }
1206                 if ($feed_data["header"]["author-nick"] != "") {
1207                         $data["nick"] = $feed_data["header"]["author-nick"];
1208                 }
1209                 if ($feed_data["header"]["author-avatar"] != "") {
1210                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1211                 }
1212                 if ($feed_data["header"]["author-id"] != "") {
1213                         $data["alias"] = $feed_data["header"]["author-id"];
1214                 }
1215                 if ($feed_data["header"]["author-location"] != "") {
1216                         $data["location"] = $feed_data["header"]["author-location"];
1217                 }
1218                 if ($feed_data["header"]["author-about"] != "") {
1219                         $data["about"] = $feed_data["header"]["author-about"];
1220                 }
1221                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1222                 // So we take the value that we just fetched, although the other one worked as well
1223                 if ($feed_data["header"]["author-link"] != "") {
1224                         $data["url"] = $feed_data["header"]["author-link"];
1225                 }
1226
1227                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1228                         $data['url'] = $data["alias"];
1229                         $data["alias"] = '';
1230                 }
1231
1232                 /// @todo Fetch location and "about" from the feed as well
1233                 return $data;
1234         }
1235
1236         /**
1237          * @brief Fetch data from a pump.io profile page
1238          *
1239          * @param string $profile_link Link to the profile page
1240          *
1241          * @return array profile data
1242          */
1243         private static function pumpioProfileData($profile_link) {
1244
1245                 $doc = new DOMDocument();
1246                 if (!@$doc->loadHTMLFile($profile_link)) {
1247                         return false;
1248                 }
1249
1250                 $xpath = new DomXPath($doc);
1251
1252                 $data = array();
1253
1254                 // This is ugly - but pump.io doesn't seem to know a better way for it
1255                 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1256                 $pos = strpos($data["name"], chr(10));
1257                 if ($pos) {
1258                         $data["name"] = trim(substr($data["name"], 0, $pos));
1259                 }
1260
1261                 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1262                 if ($avatar) {
1263                         foreach ($avatar->attributes as $attribute) {
1264                                 if ($attribute->name == "src") {
1265                                         $data["photo"] = trim($attribute->value);
1266                                 }
1267                         }
1268                 }
1269
1270                 $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
1271                 $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
1272
1273                 return $data;
1274         }
1275
1276         /**
1277          * @brief Check for pump.io contact
1278          *
1279          * @param array $webfinger Webfinger data
1280          *
1281          * @return array pump.io data
1282          */
1283         private static function pumpio($webfinger) {
1284
1285                 $data = array();
1286                 foreach ($webfinger["links"] as $link) {
1287                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1288                                 && ($link["type"] == "text/html")
1289                                 && ($link["href"] != "")
1290                         ) {
1291                                 $data["url"] = $link["href"];
1292                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1293                                 $data["notify"] = $link["href"];
1294                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1295                                 $data["poll"] = $link["href"];
1296                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1297                                 $data["dialback"] = $link["href"];
1298                         }
1299                 }
1300                 if (isset($data["poll"]) && isset($data["notify"])
1301                         && isset($data["dialback"])
1302                         && isset($data["url"])
1303                 ) {
1304                         // by now we use these fields only for the network type detection
1305                         // So we unset all data that isn't used at the moment
1306                         unset($data["dialback"]);
1307
1308                         $data["network"] = NETWORK_PUMPIO;
1309                 } else {
1310                         return false;
1311                 }
1312
1313                 $profile_data = self::pumpioProfileData($data["url"]);
1314
1315                 if (!$profile_data) {
1316                         return false;
1317                 }
1318
1319                 $data = array_merge($data, $profile_data);
1320
1321                 return $data;
1322         }
1323
1324         /**
1325          * @brief Check page for feed link
1326          *
1327          * @param string $url Page link
1328          *
1329          * @return string feed link
1330          */
1331         private static function getFeedLink($url) {
1332                 $doc = new DOMDocument();
1333
1334                 if (!@$doc->loadHTMLFile($url)) {
1335                         return false;
1336                 }
1337
1338                 $xpath = new DomXPath($doc);
1339
1340                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1341                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1342                 if (!is_object($feeds)) {
1343                         return false;
1344                 }
1345
1346                 if ($feeds->length == 0) {
1347                         return false;
1348                 }
1349
1350                 $feed_url = "";
1351
1352                 foreach ($feeds as $feed) {
1353                         $attr = array();
1354                         foreach ($feed->attributes as $attribute) {
1355                                 $attr[$attribute->name] = trim($attribute->value);
1356                         }
1357
1358                         if ($feed_url == "") {
1359                                 $feed_url = $attr["href"];
1360                         }
1361                 }
1362
1363                 return $feed_url;
1364         }
1365
1366         /**
1367          * @brief Check for feed contact
1368          *
1369          * @param string $url Profile link
1370          * @param boolean $probe Do a probe if the page contains a feed link
1371          *
1372          * @return array feed data
1373          */
1374         private static function feed($url, $probe = true) {
1375                 $ret = z_fetch_url($url);
1376                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1377                         return false;
1378                 }
1379                 $feed = $ret['body'];
1380                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1381
1382                 if (!$feed_data) {
1383                         if (!$probe) {
1384                                 return false;
1385                         }
1386
1387                         $feed_url = self::getFeedLink($url);
1388
1389                         if (!$feed_url) {
1390                                 return false;
1391                         }
1392
1393                         return self::feed($feed_url, false);
1394                 }
1395
1396                 if ($feed_data["header"]["author-name"] != "") {
1397                         $data["name"] = $feed_data["header"]["author-name"];
1398                 }
1399
1400                 if ($feed_data["header"]["author-nick"] != "") {
1401                         $data["nick"] = $feed_data["header"]["author-nick"];
1402                 }
1403
1404                 if ($feed_data["header"]["author-avatar"] != "") {
1405                         $data["photo"] = $feed_data["header"]["author-avatar"];
1406                 }
1407
1408                 if ($feed_data["header"]["author-id"] != "") {
1409                         $data["alias"] = $feed_data["header"]["author-id"];
1410                 }
1411
1412                 $data["url"] = $url;
1413                 $data["poll"] = $url;
1414
1415                 if ($feed_data["header"]["author-link"] != "") {
1416                         $data["baseurl"] = $feed_data["header"]["author-link"];
1417                 } else {
1418                         $data["baseurl"] = $data["url"];
1419                 }
1420
1421                 $data["network"] = NETWORK_FEED;
1422
1423                 return $data;
1424         }
1425
1426         /**
1427          * @brief Check for mail contact
1428          *
1429          * @param string $uri Profile link
1430          * @param integer $uid User ID
1431          *
1432          * @return array mail data
1433          */
1434         private static function mail($uri, $uid) {
1435
1436                 if (!validate_email($uri)) {
1437                         return false;
1438                 }
1439
1440                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1441
1442                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1443
1444                 if (dbm::is_result($x) && dbm::is_result($r)) {
1445                         $mailbox = construct_mailbox_name($r[0]);
1446                         $password = '';
1447                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1448                         $mbox = email_connect($mailbox, $r[0]['user'], $password);
1449                         if (!mbox) {
1450                                 return false;
1451                         }
1452                 }
1453
1454                 $msgs = email_poll($mbox, $uri);
1455                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1456
1457                 if (!count($msgs)) {
1458                         return false;
1459                 }
1460
1461                 $phost = substr($uri, strpos($uri, '@') + 1);
1462
1463                 $data = array();
1464                 $data["addr"]    = $uri;
1465                 $data["network"] = NETWORK_MAIL;
1466                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1467                 $data["nick"]    = $data["name"];
1468                 $data["photo"]   = avatar_img($uri);
1469                 $data["url"]     = 'http://'.$phost."/".$data["nick"];
1470                 $data["notify"]  = 'smtp '.random_string();
1471                 $data["poll"]    = 'email '.random_string();
1472
1473                 $x = email_msg_meta($mbox, $msgs[0]);
1474                 if (stristr($x[0]->from, $uri)) {
1475                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1476                 } elseif (stristr($x[0]->to, $uri)) {
1477                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1478                 }
1479                 if (isset($adr)) {
1480                         foreach ($adr as $feadr) {
1481                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1482                                         &&(strcasecmp($feadr->host, $phost) == 0)
1483                                         && (strlen($feadr->personal))
1484                                 ) {
1485                                         $personal = imap_mime_header_decode($feadr->personal);
1486                                         $data["name"] = "";
1487                                         foreach ($personal as $perspart) {
1488                                                 if ($perspart->charset != "default") {
1489                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1490                                                 } else {
1491                                                         $data["name"] .= $perspart->text;
1492                                                 }
1493                                         }
1494
1495                                         $data["name"] = notags($data["name"]);
1496                                 }
1497                         }
1498                 }
1499                 imap_close($mbox);
1500
1501                 return $data;
1502         }
1503
1504         /**
1505          * @brief Mix two paths together to possibly fix missing parts
1506          *
1507          * @param string $avatar Path to the avatar
1508          * @param string $base Another path that is hopefully complete
1509          *
1510          * @return string fixed avatar path
1511          */
1512         public static function fixAvatar($avatar, $base) {
1513                 $base_parts = parse_url($base);
1514
1515                 // Remove all parts that could create a problem
1516                 unset($base_parts['path']);
1517                 unset($base_parts['query']);
1518                 unset($base_parts['fragment']);
1519
1520                 $avatar_parts = parse_url($avatar);
1521
1522                 // Now we mix them
1523                 $parts = array_merge($base_parts, $avatar_parts);
1524
1525                 // And put them together again
1526                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1527                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1528                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1529                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1530                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1531                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1532
1533                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1534
1535                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1536
1537                 return $fixed;
1538         }
1539 }