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