src/Network/Probe.php

   1 <?php
   2 /**
   3  * @file src/Network/Probe.php
   4  */
   5 namespace Friendica\Network;
   6
   7 /**
   8  * @file src/Network/Probe.php
   9  * @brief Functions for probing URL
  10  */
  11
  12 use DOMDocument;
  13 use Friendica\Core\Cache;
  14 use Friendica\Core\Config;
  15 use Friendica\Core\System;
  16 use Friendica\Database\dba;
  17 use Friendica\Database\DBM;
  18 use Friendica\Model\Profile;
  19 use Friendica\Protocol\Email;
  20 use Friendica\Protocol\Feed;
  21 use Friendica\Util\Crypto;
  22 use Friendica\Util\DateTimeFormat;
  23 use Friendica\Util\Network;
  24 use Friendica\Util\XML;
  25
  26 require_once 'include/dba.php';
  27
  28 /**
  29  * @brief This class contain functions for probing URL
  30  *
  31  */
  32 class Probe
  33 {
  34         private static $baseurl;
  35
  36         /**
  37          * @brief Rearrange the array so that it always has the same order
  38          *
  39          * @param array $data Unordered data
  40          *
  41          * @return array Ordered data
  42          */
  43         private static function rearrangeData($data)
  44         {
  45                 $fields = ["name", "nick", "guid", "url", "addr", "alias",
  46                                 "photo", "community", "keywords", "location", "about",
  47                                 "batch", "notify", "poll", "request", "confirm", "poco",
  48                                 "priority", "network", "pubkey", "baseurl"];
  49
  50                 $newdata = [];
  51                 foreach ($fields as $field) {
  52                         if (isset($data[$field])) {
  53                                 $newdata[$field] = $data[$field];
  54                         } else {
  55                                 $newdata[$field] = "";
  56                         }
  57                 }
  58
  59                 // We don't use the "priority" field anymore and replace it with a dummy.
  60                 $newdata["priority"] = 0;
  61
  62                 return $newdata;
  63         }
  64
  65         /**
  66          * @brief Check if the hostname belongs to the own server
  67          *
  68          * @param string $host The hostname that is to be checked
  69          *
  70          * @return bool Does the testes hostname belongs to the own server?
  71          */
  72         private static function ownHost($host)
  73         {
  74                 $own_host = get_app()->get_hostname();
  75
  76                 $parts = parse_url($host);
  77
  78                 if (!isset($parts['scheme'])) {
  79                         $parts = parse_url('http://'.$host);
  80                 }
  81
  82                 if (!isset($parts['host'])) {
  83                         return false;
  84                 }
  85                 return $parts['host'] == $own_host;
  86         }
  87
  88         /**
  89          * @brief Probes for webfinger path via "host-meta"
  90          *
  91          * We have to check if the servers in the future still will offer this.
  92          * It seems as if it was dropped from the standard.
  93          *
  94          * @param string $host The host part of an url
  95          *
  96          * @return array with template and type of the webfinger template for JSON or XML
  97          */
  98         private static function hostMeta($host)
  99         {
 100                 // Reset the static variable
 101                 self::$baseurl = '';
 102
 103                 $ssl_url = "https://".$host."/.well-known/host-meta";
 104                 $url = "http://".$host."/.well-known/host-meta";
 105
 106                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 107                 $redirects = 0;
 108
 109                 logger("Probing for ".$host, LOGGER_DEBUG);
 110                 $xrd = null;
 111
 112                 $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 113                 if ($ret['success']) {
 114                         $xml = $ret['body'];
 115                         $xrd = XML::parseString($xml, false);
 116                         $host_url = 'https://'.$host;
 117                 }
 118
 119                 if (!is_object($xrd)) {
 120                         $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 121                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 122                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 123                                 return false;
 124                         }
 125                         $xml = $ret['body'];
 126                         $xrd = XML::parseString($xml, false);
 127                         $host_url = 'http://'.$host;
 128                 }
 129                 if (!is_object($xrd)) {
 130                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 131                         return [];
 132                 }
 133
 134                 $links = XML::elementToArray($xrd);
 135                 if (!isset($links["xrd"]["link"])) {
 136                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 137                         return [];
 138                 }
 139
 140                 $lrdd = [];
 141                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 142                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 143                 // compared to the XML webfinger. So this is commented out by now.
 144                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 145
 146                 foreach ($links["xrd"]["link"] as $value => $link) {
 147                         if (!empty($link["@attributes"])) {
 148                                 $attributes = $link["@attributes"];
 149                         } elseif ($value == "@attributes") {
 150                                 $attributes = $link;
 151                         } else {
 152                                 continue;
 153                         }
 154
 155                         if (($attributes["rel"] == "lrdd") && !empty($attributes["template"])) {
 156                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 157
 158                                 $lrdd[$type] = $attributes["template"];
 159                         }
 160                 }
 161
 162                 self::$baseurl = "http://".$host;
 163
 164                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 165
 166                 return $lrdd;
 167         }
 168
 169         /**
 170          * @brief Perform Webfinger lookup and return DFRN data
 171          *
 172          * Given an email style address, perform webfinger lookup and
 173          * return the resulting DFRN profile URL, or if no DFRN profile URL
 174          * is located, returns an OStatus subscription template (prefixed
 175          * with the string 'stat:' to identify it as on OStatus template).
 176          * If this isn't an email style address just return $webbie.
 177          * Return an empty string if email-style addresses but webfinger fails,
 178          * or if the resultant personal XRD doesn't contain a supported
 179          * subscription/friend-request attribute.
 180          *
 181          * amended 7/9/2011 to return an hcard which could save potentially loading
 182          * a lengthy content page to scrape dfrn attributes
 183          *
 184          * @param string $webbie    Address that should be probed
 185          * @param string $hcard_url Link to the hcard - is returned by reference
 186          *
 187          * @return string profile link
 188          */
 189         public static function webfingerDfrn($webbie, &$hcard_url)
 190         {
 191                 $profile_link = '';
 192
 193                 $links = self::lrdd($webbie);
 194                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 195                 if (count($links)) {
 196                         foreach ($links as $link) {
 197                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 198                                         $profile_link = $link['@attributes']['href'];
 199                                 }
 200                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 201                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 202                                 }
 203                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 204                                         $hcard_url = $link['@attributes']['href'];
 205                                 }
 206                         }
 207                 }
 208                 return $profile_link;
 209         }
 210
 211         /**
 212          * @brief Check an URI for LRDD data
 213          *
 214          * this is a replacement for the "lrdd" function.
 215          * It isn't used in this class and has some redundancies in the code.
 216          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 217          *
 218          * @param string $uri Address that should be probed
 219          *
 220          * @return array uri data
 221          */
 222         public static function lrdd($uri)
 223         {
 224                 $lrdd = self::hostMeta($uri);
 225                 $webfinger = null;
 226
 227                 if (is_bool($lrdd)) {
 228                         return [];
 229                 }
 230
 231                 if (!$lrdd) {
 232                         $parts = @parse_url($uri);
 233                         if (!$parts) {
 234                                 return [];
 235                         }
 236
 237                         $host = $parts["host"];
 238                         if (!empty($parts["port"])) {
 239                                 $host .= ':'.$parts["port"];
 240                         }
 241
 242                         $path_parts = explode("/", trim($parts["path"], "/"));
 243
 244                         $nick = array_pop($path_parts);
 245
 246                         do {
 247                                 $lrdd = self::hostMeta($host);
 248                                 $host .= "/".array_shift($path_parts);
 249                         } while (!$lrdd && (sizeof($path_parts) > 0));
 250                 }
 251
 252                 if (!$lrdd) {
 253                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 254                         return [];
 255                 }
 256
 257                 foreach ($lrdd as $type => $template) {
 258                         if ($webfinger) {
 259                                 continue;
 260                         }
 261
 262                         $path = str_replace('{uri}', urlencode($uri), $template);
 263                         $webfinger = self::webfinger($path, $type);
 264
 265                         if (!$webfinger && (strstr($uri, "@"))) {
 266                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 267                                 $webfinger = self::webfinger($path, $type);
 268                         }
 269
 270                         // Special treatment for Mastodon
 271                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 272                         // But the webfinger for this format fails.
 273                         if (!$webfinger && !empty($nick)) {
 274                                 // Mastodon uses a "@" as prefix for usernames in their url format
 275                                 $nick = ltrim($nick, '@');
 276
 277                                 $addr = $nick."@".$host;
 278
 279                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 280                                 $webfinger = self::webfinger($path, $type);
 281                         }
 282                 }
 283
 284                 if (!is_array($webfinger["links"])) {
 285                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 286                         return false;
 287                 }
 288
 289                 $data = [];
 290
 291                 foreach ($webfinger["links"] as $link) {
 292                         $data[] = ["@attributes" => $link];
 293                 }
 294
 295                 if (is_array($webfinger["aliases"])) {
 296                         foreach ($webfinger["aliases"] as $alias) {
 297                                 $data[] = ["@attributes" =>
 298                                                         ["rel" => "alias",
 299                                                                 "href" => $alias]];
 300                         }
 301                 }
 302
 303                 return $data;
 304         }
 305
 306         /**
 307          * @brief Fetch information (protocol endpoints and user information) about a given uri
 308          *
 309          * @param string  $uri     Address that should be probed
 310          * @param string  $network Test for this specific network
 311          * @param integer $uid     User ID for the probe (only used for mails)
 312          * @param boolean $cache   Use cached values?
 313          *
 314          * @return array uri data
 315          */
 316         public static function uri($uri, $network = "", $uid = -1, $cache = true)
 317         {
 318                 if ($cache) {
 319                         $result = Cache::get("Probe::uri:".$network.":".$uri);
 320                         if (!is_null($result)) {
 321                                 return $result;
 322                         }
 323                 }
 324
 325                 if ($uid == -1) {
 326                         $uid = local_user();
 327                 }
 328
 329                 $data = self::detect($uri, $network, $uid);
 330
 331                 if (!isset($data["url"])) {
 332                         $data["url"] = $uri;
 333                 }
 334
 335                 if (x($data, "photo")) {
 336                         $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
 337                 } else {
 338                         $data["photo"] = System::baseUrl().'/images/person-175.jpg';
 339                 }
 340
 341                 if (empty($data["name"])) {
 342                         if (!empty($data["nick"])) {
 343                                 $data["name"] = $data["nick"];
 344                         }
 345
 346                         if (!x($data, "name")) {
 347                                 $data["name"] = $data["url"];
 348                         }
 349                 }
 350
 351                 if (empty($data["nick"])) {
 352                         $data["nick"] = strtolower($data["name"]);
 353
 354                         if (strpos($data['nick'], ' ')) {
 355                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 356                         }
 357                 }
 358
 359                 if (!empty(self::$baseurl)) {
 360                         $data["baseurl"] = self::$baseurl;
 361                 }
 362
 363                 if (empty($data["network"])) {
 364                         $data["network"] = NETWORK_PHANTOM;
 365                 }
 366
 367                 $data = self::rearrangeData($data);
 368
 369                 // Only store into the cache if the value seems to be valid
 370                 if (!in_array($data['network'], [NETWORK_PHANTOM, NETWORK_MAIL])) {
 371                         Cache::set("Probe::uri:".$network.":".$uri, $data, CACHE_DAY);
 372
 373                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 374                         /// The biggest problem is the avatar picture that could have a reduced image size.
 375                         /// It should only be updated if the existing picture isn't existing anymore.
 376                         /// We only update the contact when it is no probing for a specific network.
 377                         if (($data['network'] != NETWORK_FEED)
 378                                 && ($network == "")
 379                                 && $data["name"]
 380                                 && $data["nick"]
 381                                 && $data["url"]
 382                                 && $data["addr"]
 383                                 && $data["poll"]
 384                         ) {
 385                                 $fields = ['name' => $data['name'],
 386                                                 'nick' => $data['nick'],
 387                                                 'url' => $data['url'],
 388                                                 'addr' => $data['addr'],
 389                                                 'photo' => $data['photo'],
 390                                                 'keywords' => $data['keywords'],
 391                                                 'location' => $data['location'],
 392                                                 'about' => $data['about'],
 393                                                 'notify' => $data['notify'],
 394                                                 'network' => $data['network'],
 395                                                 'server_url' => $data['baseurl']];
 396
 397                                 // This doesn't cover the case when a community isn't a community anymore
 398                                 if (!empty($data['community']) && $data['community']) {
 399                                         $fields['community'] = $data['community'];
 400                                         $fields['contact-type'] = ACCOUNT_TYPE_COMMUNITY;
 401                                 }
 402
 403                                 $fieldnames = [];
 404
 405                                 foreach ($fields as $key => $val) {
 406                                         if (empty($val)) {
 407                                                 unset($fields[$key]);
 408                                         } else {
 409                                                 $fieldnames[] = $key;
 410                                         }
 411                                 }
 412
 413                                 $fields['updated'] = DBM::date();
 414
 415                                 $condition = ['nurl' => normalise_link($data["url"])];
 416
 417                                 $old_fields = dba::selectFirst('gcontact', $fieldnames, $condition);
 418
 419                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 420                                 // In difference to the public contacts we want to have every contact
 421                                 // in the world in our global contacts.
 422                                 if (!$old_fields) {
 423                                         $old_fields = true;
 424
 425                                         // These values have to be set only on insert
 426                                         $fields['photo'] = $data['photo'];
 427                                         $fields['created'] = DateTimeFormat::utcNow();
 428                                 }
 429
 430                                 dba::update('gcontact', $fields, $condition, $old_fields);
 431
 432                                 $fields = ['name' => $data['name'],
 433                                                 'nick' => $data['nick'],
 434                                                 'url' => $data['url'],
 435                                                 'addr' => $data['addr'],
 436                                                 'alias' => $data['alias'],
 437                                                 'keywords' => $data['keywords'],
 438                                                 'location' => $data['location'],
 439                                                 'about' => $data['about'],
 440                                                 'batch' => $data['batch'],
 441                                                 'notify' => $data['notify'],
 442                                                 'poll' => $data['poll'],
 443                                                 'request' => $data['request'],
 444                                                 'confirm' => $data['confirm'],
 445                                                 'poco' => $data['poco'],
 446                                                 'network' => $data['network'],
 447                                                 'pubkey' => $data['pubkey'],
 448                                                 'priority' => $data['priority'],
 449                                                 'writable' => true,
 450                                                 'rel' => CONTACT_IS_SHARING];
 451
 452                                 $fieldnames = [];
 453
 454                                 foreach ($fields as $key => $val) {
 455                                         if (empty($val)) {
 456                                                 unset($fields[$key]);
 457                                         } else {
 458                                                 $fieldnames[] = $key;
 459                                         }
 460                                 }
 461
 462                                 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 463
 464                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 465                                 // This won't trigger an insert. This is intended, since we only need
 466                                 // public contacts for everyone we store items from.
 467                                 // We don't need to store every contact on the planet.
 468                                 $old_fields = dba::selectFirst('contact', $fieldnames, $condition);
 469
 470                                 $fields['name-date'] = DateTimeFormat::utcNow();
 471                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 472                                 $fields['success_update'] = DateTimeFormat::utcNow();
 473
 474                                 dba::update('contact', $fields, $condition, $old_fields);
 475                         }
 476                 }
 477
 478                 return $data;
 479         }
 480
 481         /**
 482          * @brief Switch the scheme of an url between http and https
 483          *
 484          * @param string $url URL
 485          *
 486          * @return string switched URL
 487          */
 488         private static function switchScheme($url)
 489         {
 490                 $parts = parse_url($url);
 491
 492                 if (!isset($parts['scheme'])) {
 493                         return $url;
 494                 }
 495
 496                 if ($parts['scheme'] == 'http') {
 497                         $url = str_replace('http://', 'https://', $url);
 498                 } elseif ($parts['scheme'] == 'https') {
 499                         $url = str_replace('https://', 'http://', $url);
 500                 }
 501
 502                 return $url;
 503         }
 504
 505         /**
 506          * @brief Checks if a profile url should be OStatus but only provides partial information
 507          *
 508          * @param array  $webfinger Webfinger data
 509          * @param string $lrdd      Path template for webfinger request
 510          * @param string $type      type
 511          *
 512          * @return array fixed webfinger data
 513          */
 514         private static function fixOStatus($webfinger, $lrdd, $type)
 515         {
 516                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 517                         return $webfinger;
 518                 }
 519
 520                 $is_ostatus = false;
 521                 $has_key = false;
 522
 523                 foreach ($webfinger['links'] as $link) {
 524                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 525                                 $is_ostatus = true;
 526                         }
 527                         if ($link['rel'] == 'magic-public-key') {
 528                                 $has_key = true;
 529                         }
 530                 }
 531
 532                 if (!$is_ostatus || $has_key) {
 533                         return $webfinger;
 534                 }
 535
 536                 $url = self::switchScheme($webfinger['subject']);
 537                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 538                 $webfinger2 = self::webfinger($path, $type);
 539
 540                 // Is the new webfinger detectable as OStatus?
 541                 if (self::ostatus($webfinger2, true)) {
 542                         $webfinger = $webfinger2;
 543                 }
 544
 545                 return $webfinger;
 546         }
 547
 548         /**
 549          * @brief Fetch information (protocol endpoints and user information) about a given uri
 550          *
 551          * This function is only called by the "uri" function that adds caching and rearranging of data.
 552          *
 553          * @param string  $uri     Address that should be probed
 554          * @param string  $network Test for this specific network
 555          * @param integer $uid     User ID for the probe (only used for mails)
 556          *
 557          * @return array uri data
 558          */
 559         private static function detect($uri, $network, $uid)
 560         {
 561                 $parts = parse_url($uri);
 562
 563                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 564                         $host = $parts["host"];
 565                         if (!empty($parts["port"])) {
 566                                 $host .= ':'.$parts["port"];
 567                         }
 568
 569                         if ($host == 'twitter.com') {
 570                                 return ["network" => NETWORK_TWITTER];
 571                         }
 572                         $lrdd = self::hostMeta($host);
 573
 574                         if (is_bool($lrdd)) {
 575                                 return [];
 576                         }
 577
 578                         $path_parts = explode("/", trim($parts["path"], "/"));
 579
 580                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 581                                 $host .= "/".array_shift($path_parts);
 582                                 $lrdd = self::hostMeta($host);
 583                         }
 584                         if (!$lrdd) {
 585                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 586                                 return self::feed($uri);
 587                         }
 588                         $nick = array_pop($path_parts);
 589
 590                         // Mastodon uses a "@" as prefix for usernames in their url format
 591                         $nick = ltrim($nick, '@');
 592
 593                         $addr = $nick."@".$host;
 594                 } elseif (strstr($uri, '@')) {
 595                         // If the URI starts with "mailto:" then jump directly to the mail detection
 596                         if (strpos($uri, 'mailto:') !== false) {
 597                                 $uri = str_replace('mailto:', '', $uri);
 598                                 return self::mail($uri, $uid);
 599                         }
 600
 601                         if ($network == NETWORK_MAIL) {
 602                                 return self::mail($uri, $uid);
 603                         }
 604                         // Remove "acct:" from the URI
 605                         $uri = str_replace('acct:', '', $uri);
 606
 607                         $host = substr($uri, strpos($uri, '@') + 1);
 608                         $nick = substr($uri, 0, strpos($uri, '@'));
 609
 610                         if (strpos($uri, '@twitter.com')) {
 611                                 return ["network" => NETWORK_TWITTER];
 612                         }
 613                         $lrdd = self::hostMeta($host);
 614
 615                         if (is_bool($lrdd)) {
 616                                 return [];
 617                         }
 618
 619                         if (!$lrdd) {
 620                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 621                                 return self::mail($uri, $uid);
 622                         }
 623                         $addr = $uri;
 624                 } else {
 625                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 626                         return false;
 627                 }
 628
 629                 $webfinger = false;
 630
 631                 /// @todo Do we need the prefix "acct:" or "acct://"?
 632
 633                 foreach ($lrdd as $type => $template) {
 634                         if ($webfinger) {
 635                                 continue;
 636                         }
 637
 638                         // At first try it with the given uri
 639                         $path = str_replace('{uri}', urlencode($uri), $template);
 640                         $webfinger = self::webfinger($path, $type);
 641
 642                         // Fix possible problems with GNU Social probing to wrong scheme
 643                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 644
 645                         // We cannot be sure that the detected address was correct, so we don't use the values
 646                         if ($webfinger && ($uri != $addr)) {
 647                                 $nick = "";
 648                                 $addr = "";
 649                         }
 650
 651                         // Try webfinger with the address (user@domain.tld)
 652                         if (!$webfinger) {
 653                                 $path = str_replace('{uri}', urlencode($addr), $template);
 654                                 $webfinger = self::webfinger($path, $type);
 655                         }
 656
 657                         // Mastodon needs to have it with "acct:"
 658                         if (!$webfinger) {
 659                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 660                                 $webfinger = self::webfinger($path, $type);
 661                         }
 662                 }
 663
 664                 if (!$webfinger) {
 665                         return self::feed($uri);
 666                 }
 667
 668                 $result = false;
 669
 670                 logger("Probing ".$uri, LOGGER_DEBUG);
 671
 672                 if (in_array($network, ["", NETWORK_DFRN])) {
 673                         $result = self::dfrn($webfinger);
 674                 }
 675                 if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
 676                         $result = self::diaspora($webfinger);
 677                 }
 678                 if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
 679                         $result = self::ostatus($webfinger);
 680                 }
 681                 if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
 682                         $result = self::pumpio($webfinger, $addr);
 683                 }
 684                 if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
 685                         $result = self::feed($uri);
 686                 } else {
 687                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 688                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 689                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 690                                 $result["nick"] = $nick;
 691                         }
 692
 693                         if (empty($result["addr"]) && ($addr != "")) {
 694                                 $result["addr"] = $addr;
 695                         }
 696                 }
 697
 698                 if (empty($result["network"])) {
 699                         $result["network"] = NETWORK_PHANTOM;
 700                 }
 701
 702                 if (empty($result["url"])) {
 703                         $result["url"] = $uri;
 704                 }
 705
 706                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 707
 708                 if (empty($result["baseurl"])) {
 709                         $pos = strpos($result["url"], $host);
 710                         if ($pos) {
 711                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 712                         }
 713                 }
 714                 return $result;
 715         }
 716
 717         /**
 718          * @brief Perform a webfinger request.
 719          *
 720          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 721          *
 722          * @param string $url  Address that should be probed
 723          * @param string $type type
 724          *
 725          * @return array webfinger data
 726          */
 727         private static function webfinger($url, $type)
 728         {
 729                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 730                 $redirects = 0;
 731
 732                 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 733                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 734                         return false;
 735                 }
 736                 $data = $ret['body'];
 737
 738                 $webfinger = json_decode($data, true);
 739                 if (is_array($webfinger)) {
 740                         if (!isset($webfinger["links"])) {
 741                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 742                                 return false;
 743                         }
 744                         return $webfinger;
 745                 }
 746
 747                 // If it is not JSON, maybe it is XML
 748                 $xrd = XML::parseString($data, false);
 749                 if (!is_object($xrd)) {
 750                         logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
 751                         return false;
 752                 }
 753
 754                 $xrd_arr = XML::elementToArray($xrd);
 755                 if (!isset($xrd_arr["xrd"]["link"])) {
 756                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 757                         return false;
 758                 }
 759
 760                 $webfinger = [];
 761
 762                 if (!empty($xrd_arr["xrd"]["subject"])) {
 763                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 764                 }
 765
 766                 if (!empty($xrd_arr["xrd"]["alias"])) {
 767                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 768                 }
 769
 770                 $webfinger["links"] = [];
 771
 772                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 773                         if (!empty($data["@attributes"])) {
 774                                 $attributes = $data["@attributes"];
 775                         } elseif ($value == "@attributes") {
 776                                 $attributes = $data;
 777                         } else {
 778                                 continue;
 779                         }
 780
 781                         $webfinger["links"][] = $attributes;
 782                 }
 783                 return $webfinger;
 784         }
 785
 786         /**
 787          * @brief Poll the Friendica specific noscrape page.
 788          *
 789          * "noscrape" is a faster alternative to fetch the data from the hcard.
 790          * This functionality was originally created for the directory.
 791          *
 792          * @param string $noscrape_url Link to the noscrape page
 793          * @param array  $data         The already fetched data
 794          *
 795          * @return array noscrape data
 796          */
 797         private static function pollNoscrape($noscrape_url, $data)
 798         {
 799                 $ret = Network::curl($noscrape_url);
 800                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 801                         return false;
 802                 }
 803                 $content = $ret['body'];
 804                 if (!$content) {
 805                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 806                         return false;
 807                 }
 808
 809                 $json = json_decode($content, true);
 810                 if (!is_array($json)) {
 811                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 812                         return false;
 813                 }
 814
 815                 if (!empty($json["fn"])) {
 816                         $data["name"] = $json["fn"];
 817                 }
 818
 819                 if (!empty($json["addr"])) {
 820                         $data["addr"] = $json["addr"];
 821                 }
 822
 823                 if (!empty($json["nick"])) {
 824                         $data["nick"] = $json["nick"];
 825                 }
 826
 827                 if (!empty($json["guid"])) {
 828                         $data["guid"] = $json["guid"];
 829                 }
 830
 831                 if (!empty($json["comm"])) {
 832                         $data["community"] = $json["comm"];
 833                 }
 834
 835                 if (!empty($json["tags"])) {
 836                         $keywords = implode(" ", $json["tags"]);
 837                         if ($keywords != "") {
 838                                 $data["keywords"] = $keywords;
 839                         }
 840                 }
 841
 842                 $location = Profile::formatLocation($json);
 843                 if ($location) {
 844                         $data["location"] = $location;
 845                 }
 846
 847                 if (!empty($json["about"])) {
 848                         $data["about"] = $json["about"];
 849                 }
 850
 851                 if (!empty($json["key"])) {
 852                         $data["pubkey"] = $json["key"];
 853                 }
 854
 855                 if (!empty($json["photo"])) {
 856                         $data["photo"] = $json["photo"];
 857                 }
 858
 859                 if (!empty($json["dfrn-request"])) {
 860                         $data["request"] = $json["dfrn-request"];
 861                 }
 862
 863                 if (!empty($json["dfrn-confirm"])) {
 864                         $data["confirm"] = $json["dfrn-confirm"];
 865                 }
 866
 867                 if (!empty($json["dfrn-notify"])) {
 868                         $data["notify"] = $json["dfrn-notify"];
 869                 }
 870
 871                 if (!empty($json["dfrn-poll"])) {
 872                         $data["poll"] = $json["dfrn-poll"];
 873                 }
 874
 875                 return $data;
 876         }
 877
 878         /**
 879          * @brief Check for valid DFRN data
 880          *
 881          * @param array $data DFRN data
 882          *
 883          * @return int Number of errors
 884          */
 885         public static function validDfrn($data)
 886         {
 887                 $errors = 0;
 888                 if (!isset($data['key'])) {
 889                         $errors ++;
 890                 }
 891                 if (!isset($data['dfrn-request'])) {
 892                         $errors ++;
 893                 }
 894                 if (!isset($data['dfrn-confirm'])) {
 895                         $errors ++;
 896                 }
 897                 if (!isset($data['dfrn-notify'])) {
 898                         $errors ++;
 899                 }
 900                 if (!isset($data['dfrn-poll'])) {
 901                         $errors ++;
 902                 }
 903                 return $errors;
 904         }
 905
 906         /**
 907          * @brief Fetch data from a DFRN profile page and via "noscrape"
 908          *
 909          * @param string $profile_link Link to the profile page
 910          *
 911          * @return array profile data
 912          */
 913         public static function profile($profile_link)
 914         {
 915                 $data = [];
 916
 917                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 918
 919                 // Fetch data via noscrape - this is faster
 920                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 921                 $data = self::pollNoscrape($noscrape_url, $data);
 922
 923                 if (!isset($data["notify"])
 924                         || !isset($data["confirm"])
 925                         || !isset($data["request"])
 926                         || !isset($data["poll"])
 927                         || !isset($data["poco"])
 928                         || !isset($data["name"])
 929                         || !isset($data["photo"])
 930                 ) {
 931                         $data = self::pollHcard($profile_link, $data, true);
 932                 }
 933
 934                 $prof_data = [];
 935                 $prof_data["addr"]         = $data["addr"];
 936                 $prof_data["nick"]         = $data["nick"];
 937                 $prof_data["dfrn-request"] = $data["request"];
 938                 $prof_data["dfrn-confirm"] = $data["confirm"];
 939                 $prof_data["dfrn-notify"]  = $data["notify"];
 940                 $prof_data["dfrn-poll"]    = $data["poll"];
 941                 $prof_data["dfrn-poco"]    = $data["poco"];
 942                 $prof_data["photo"]        = $data["photo"];
 943                 $prof_data["fn"]           = $data["name"];
 944                 $prof_data["key"]          = $data["pubkey"];
 945
 946                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 947
 948                 return $prof_data;
 949         }
 950
 951         /**
 952          * @brief Check for DFRN contact
 953          *
 954          * @param array $webfinger Webfinger data
 955          *
 956          * @return array DFRN data
 957          */
 958         private static function dfrn($webfinger)
 959         {
 960                 $hcard_url = "";
 961                 $data = [];
 962                 foreach ($webfinger["links"] as $link) {
 963                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 964                                 $data["network"] = NETWORK_DFRN;
 965                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 966                                 $data["poll"] = $link["href"];
 967                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 968                                 $data["url"] = $link["href"];
 969                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 970                                 $hcard_url = $link["href"];
 971                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 972                                 $data["poco"] = $link["href"];
 973                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 974                                 $data["photo"] = $link["href"];
 975                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 976                                 $data["baseurl"] = trim($link["href"], '/');
 977                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 978                                 $data["guid"] = $link["href"];
 979                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 980                                 $data["pubkey"] = base64_decode($link["href"]);
 981
 982                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 983                                 if (strstr($data["pubkey"], 'RSA ')) {
 984                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 985                                 }
 986                         }
 987                 }
 988
 989                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 990                         foreach ($webfinger["aliases"] as $alias) {
 991                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
 992                                         $data["alias"] = $alias;
 993                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 994                                         $data["addr"] = substr($alias, 5);
 995                                 }
 996                         }
 997                 }
 998
 999                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1000                         $data["addr"] = substr($webfinger["subject"], 5);
1001                 }
1002
1003                 if (!isset($data["network"]) || ($hcard_url == "")) {
1004                         return false;
1005                 }
1006
1007                 // Fetch data via noscrape - this is faster
1008                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1009                 $data = self::pollNoscrape($noscrape_url, $data);
1010
1011                 if (isset($data["notify"])
1012                         && isset($data["confirm"])
1013                         && isset($data["request"])
1014                         && isset($data["poll"])
1015                         && isset($data["name"])
1016                         && isset($data["photo"])
1017                 ) {
1018                         return $data;
1019                 }
1020
1021                 $data = self::pollHcard($hcard_url, $data, true);
1022
1023                 return $data;
1024         }
1025
1026         /**
1027          * @brief Poll the hcard page (Diaspora and Friendica specific)
1028          *
1029          * @param string  $hcard_url Link to the hcard page
1030          * @param array   $data      The already fetched data
1031          * @param boolean $dfrn      Poll DFRN specific data
1032          *
1033          * @return array hcard data
1034          */
1035         private static function pollHcard($hcard_url, $data, $dfrn = false)
1036         {
1037                 $ret = Network::curl($hcard_url);
1038                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1039                         return false;
1040                 }
1041                 $content = $ret['body'];
1042                 if (!$content) {
1043                         return false;
1044                 }
1045
1046                 $doc = new DOMDocument();
1047                 if (!@$doc->loadHTML($content)) {
1048                         return false;
1049                 }
1050
1051                 $xpath = new DomXPath($doc);
1052
1053                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1054                 if (!is_object($vcards)) {
1055                         return false;
1056                 }
1057
1058                 if (!isset($data["baseurl"])) {
1059                         $data["baseurl"] = "";
1060                 }
1061
1062                 if ($vcards->length > 0) {
1063                         $vcard = $vcards->item(0);
1064
1065                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1066                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1067                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1068                         if (($search->length > 0) && empty($data["guid"])) {
1069                                 $data["guid"] = $search->item(0)->nodeValue;
1070                         }
1071
1072                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1073                         if ($search->length > 0) {
1074                                 $data["nick"] = $search->item(0)->nodeValue;
1075                         }
1076
1077                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1078                         if ($search->length > 0) {
1079                                 $data["name"] = $search->item(0)->nodeValue;
1080                         }
1081
1082                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1083                         if ($search->length > 0) {
1084                                 $data["searchable"] = $search->item(0)->nodeValue;
1085                         }
1086
1087                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1088                         if ($search->length > 0) {
1089                                 $data["pubkey"] = $search->item(0)->nodeValue;
1090                                 if (strstr($data["pubkey"], 'RSA ')) {
1091                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1092                                 }
1093                         }
1094
1095                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1096                         if ($search->length > 0) {
1097                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1098                         }
1099                 }
1100
1101                 $avatar = [];
1102                 if (!empty($vcard)) {
1103                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1104                         foreach ($photos as $photo) {
1105                                 $attr = [];
1106                                 foreach ($photo->attributes as $attribute) {
1107                                         $attr[$attribute->name] = trim($attribute->value);
1108                                 }
1109
1110                                 if (isset($attr["src"]) && isset($attr["width"])) {
1111                                         $avatar[$attr["width"]] = $attr["src"];
1112                                 }
1113
1114                                 // We don't have a width. So we just take everything that we got.
1115                                 // This is a Hubzilla workaround which doesn't send a width.
1116                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1117                                         $avatar[] = $attr["src"];
1118                                 }
1119                         }
1120                 }
1121
1122                 if (sizeof($avatar)) {
1123                         ksort($avatar);
1124                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1125                 }
1126
1127                 if ($dfrn) {
1128                         // Poll DFRN specific data
1129                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1130                         if ($search->length > 0) {
1131                                 foreach ($search as $link) {
1132                                         //$data["request"] = $search->item(0)->nodeValue;
1133                                         $attr = [];
1134                                         foreach ($link->attributes as $attribute) {
1135                                                 $attr[$attribute->name] = trim($attribute->value);
1136                                         }
1137
1138                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1139                                 }
1140                         }
1141
1142                         // Older Friendica versions had used the "uid" field differently than newer versions
1143                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1144                                 unset($data["guid"]);
1145                         }
1146                 }
1147
1148
1149                 return $data;
1150         }
1151
1152         /**
1153          * @brief Check for Diaspora contact
1154          *
1155          * @param array $webfinger Webfinger data
1156          *
1157          * @return array Diaspora data
1158          */
1159         private static function diaspora($webfinger)
1160         {
1161                 $hcard_url = "";
1162                 $data = [];
1163                 foreach ($webfinger["links"] as $link) {
1164                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1165                                 $hcard_url = $link["href"];
1166                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1167                                 $data["baseurl"] = trim($link["href"], '/');
1168                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1169                                 $data["guid"] = $link["href"];
1170                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1171                                 $data["url"] = $link["href"];
1172                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1173                                 $data["poll"] = $link["href"];
1174                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1175                                 $data["poco"] = $link["href"];
1176                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1177                                 $data["notify"] = $link["href"];
1178                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1179                                 $data["pubkey"] = base64_decode($link["href"]);
1180
1181                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1182                                 if (strstr($data["pubkey"], 'RSA ')) {
1183                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1184                                 }
1185                         }
1186                 }
1187
1188                 if (!isset($data["url"]) || ($hcard_url == "")) {
1189                         return false;
1190                 }
1191
1192                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1193                         foreach ($webfinger["aliases"] as $alias) {
1194                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1195                                         $data["alias"] = $alias;
1196                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1197                                         $data["addr"] = substr($alias, 5);
1198                                 }
1199                         }
1200                 }
1201
1202                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1203                         $data["addr"] = substr($webfinger["subject"], 5);
1204                 }
1205
1206                 // Fetch further information from the hcard
1207                 $data = self::pollHcard($hcard_url, $data);
1208
1209                 if (!$data) {
1210                         return false;
1211                 }
1212
1213                 if (isset($data["url"])
1214                         && isset($data["guid"])
1215                         && isset($data["baseurl"])
1216                         && isset($data["pubkey"])
1217                         && ($hcard_url != "")
1218                 ) {
1219                         $data["network"] = NETWORK_DIASPORA;
1220
1221                         // The Diaspora handle must always be lowercase
1222                         if (!empty($data["addr"])) {
1223                                 $data["addr"] = strtolower($data["addr"]);
1224                         }
1225
1226                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1227                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1228                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1229                 } else {
1230                         return false;
1231                 }
1232
1233                 return $data;
1234         }
1235
1236         /**
1237          * @brief Check for OStatus contact
1238          *
1239          * @param array $webfinger Webfinger data
1240          * @param bool  $short     Short detection mode
1241          *
1242          * @return array|bool OStatus data or "false" on error or "true" on short mode
1243          */
1244         private static function ostatus($webfinger, $short = false)
1245         {
1246                 $data = [];
1247
1248                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1249                         foreach ($webfinger["aliases"] as $alias) {
1250                                 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1251                                         $data["addr"] = str_replace('acct:', '', $alias);
1252                                 }
1253                         }
1254                 }
1255
1256                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1257                         && !strstr(normalise_link($webfinger["subject"]), "http://")
1258                 ) {
1259                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1260                 }
1261
1262                 $pubkey = "";
1263                 if (is_array($webfinger["links"])) {
1264                         foreach ($webfinger["links"] as $link) {
1265                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1266                                         && ($link["type"] == "text/html")
1267                                         && ($link["href"] != "")
1268                                 ) {
1269                                         $data["url"] = $link["href"];
1270                                 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1271                                         $data["notify"] = $link["href"];
1272                                 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1273                                         $data["poll"] = $link["href"];
1274                                 } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1275                                         $pubkey = $link["href"];
1276
1277                                         if (substr($pubkey, 0, 5) === 'data:') {
1278                                                 if (strstr($pubkey, ',')) {
1279                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1280                                                 } else {
1281                                                         $pubkey = substr($pubkey, 5);
1282                                                 }
1283                                         } elseif (normalise_link($pubkey) == 'http://') {
1284                                                 $ret = Network::curl($pubkey);
1285                                                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1286                                                         return false;
1287                                                 }
1288                                                 $pubkey = $ret['body'];
1289                                         }
1290
1291                                         $key = explode(".", $pubkey);
1292
1293                                         if (sizeof($key) >= 3) {
1294                                                 $m = base64url_decode($key[1]);
1295                                                 $e = base64url_decode($key[2]);
1296                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1297                                         }
1298                                 }
1299                         }
1300                 }
1301
1302                 if (isset($data["notify"]) && isset($data["pubkey"])
1303                         && isset($data["poll"])
1304                         && isset($data["url"])
1305                 ) {
1306                         $data["network"] = NETWORK_OSTATUS;
1307                 } else {
1308                         return false;
1309                 }
1310
1311                 if ($short) {
1312                         return true;
1313                 }
1314
1315                 // Fetch all additional data from the feed
1316                 $ret = Network::curl($data["poll"]);
1317                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1318                         return false;
1319                 }
1320                 $feed = $ret['body'];
1321                 $dummy1 = null;
1322                 $dummy2 = null;
1323                 $dummy2 = null;
1324                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1325                 if (!$feed_data) {
1326                         return false;
1327                 }
1328
1329                 if (!empty($feed_data["header"]["author-name"])) {
1330                         $data["name"] = $feed_data["header"]["author-name"];
1331                 }
1332                 if (!empty($feed_data["header"]["author-nick"])) {
1333                         $data["nick"] = $feed_data["header"]["author-nick"];
1334                 }
1335                 if (!empty($feed_data["header"]["author-avatar"])) {
1336                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1337                 }
1338                 if (!empty($feed_data["header"]["author-id"])) {
1339                         $data["alias"] = $feed_data["header"]["author-id"];
1340                 }
1341                 if (!empty($feed_data["header"]["author-location"])) {
1342                         $data["location"] = $feed_data["header"]["author-location"];
1343                 }
1344                 if (!empty($feed_data["header"]["author-about"])) {
1345                         $data["about"] = $feed_data["header"]["author-about"];
1346                 }
1347                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1348                 // So we take the value that we just fetched, although the other one worked as well
1349                 if (!empty($feed_data["header"]["author-link"])) {
1350                         $data["url"] = $feed_data["header"]["author-link"];
1351                 }
1352
1353                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1354                         $data['url'] = $data["alias"];
1355                         $data["alias"] = '';
1356                 }
1357
1358                 /// @todo Fetch location and "about" from the feed as well
1359                 return $data;
1360         }
1361
1362         /**
1363          * @brief Fetch data from a pump.io profile page
1364          *
1365          * @param string $profile_link Link to the profile page
1366          *
1367          * @return array profile data
1368          */
1369         private static function pumpioProfileData($profile_link)
1370         {
1371                 $doc = new DOMDocument();
1372                 if (!@$doc->loadHTMLFile($profile_link)) {
1373                         return false;
1374                 }
1375
1376                 $xpath = new DomXPath($doc);
1377
1378                 $data = [];
1379
1380                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1381
1382                 if ($data["name"] == '') {
1383                         // This is ugly - but pump.io doesn't seem to know a better way for it
1384                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1385                         $pos = strpos($data["name"], chr(10));
1386                         if ($pos) {
1387                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1388                         }
1389                 }
1390
1391                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1392
1393                 if ($data["location"] == '') {
1394                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1395                 }
1396
1397                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1398
1399                 if ($data["about"] == '') {
1400                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1401                 }
1402
1403                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1404                 if (!$avatar) {
1405                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1406                 }
1407                 if ($avatar) {
1408                         foreach ($avatar->attributes as $attribute) {
1409                                 if ($attribute->name == "src") {
1410                                         $data["photo"] = trim($attribute->value);
1411                                 }
1412                         }
1413                 }
1414
1415                 return $data;
1416         }
1417
1418         /**
1419          * @brief Check for pump.io contact
1420          *
1421          * @param array $webfinger Webfinger data
1422          *
1423          * @return array pump.io data
1424          */
1425         private static function pumpio($webfinger, $addr)
1426         {
1427                 $data = [];
1428                 foreach ($webfinger["links"] as $link) {
1429                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1430                                 && ($link["type"] == "text/html")
1431                                 && ($link["href"] != "")
1432                         ) {
1433                                 $data["url"] = $link["href"];
1434                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1435                                 $data["notify"] = $link["href"];
1436                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1437                                 $data["poll"] = $link["href"];
1438                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1439                                 $data["dialback"] = $link["href"];
1440                         }
1441                 }
1442                 if (isset($data["poll"]) && isset($data["notify"])
1443                         && isset($data["dialback"])
1444                         && isset($data["url"])
1445                 ) {
1446                         // by now we use these fields only for the network type detection
1447                         // So we unset all data that isn't used at the moment
1448                         unset($data["dialback"]);
1449
1450                         $data["network"] = NETWORK_PUMPIO;
1451                 } else {
1452                         return false;
1453                 }
1454
1455                 $profile_data = self::pumpioProfileData($data["url"]);
1456
1457                 if (!$profile_data) {
1458                         return false;
1459                 }
1460
1461                 $data = array_merge($data, $profile_data);
1462
1463                 if (($addr != '') && ($data['name'] != '')) {
1464                         $name = trim(str_replace($addr, '', $data['name']));
1465                         if ($name != '') {
1466                                 $data['name'] = $name;
1467                         }
1468                 }
1469
1470                 return $data;
1471         }
1472
1473         /**
1474          * @brief Check page for feed link
1475          *
1476          * @param string $url Page link
1477          *
1478          * @return string feed link
1479          */
1480         private static function getFeedLink($url)
1481         {
1482                 $doc = new DOMDocument();
1483
1484                 if (!@$doc->loadHTMLFile($url)) {
1485                         return false;
1486                 }
1487
1488                 $xpath = new DomXPath($doc);
1489
1490                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1491                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1492                 if (!is_object($feeds)) {
1493                         return false;
1494                 }
1495
1496                 if ($feeds->length == 0) {
1497                         return false;
1498                 }
1499
1500                 $feed_url = "";
1501
1502                 foreach ($feeds as $feed) {
1503                         $attr = [];
1504                         foreach ($feed->attributes as $attribute) {
1505                                 $attr[$attribute->name] = trim($attribute->value);
1506                         }
1507
1508                         if ($feed_url == "") {
1509                                 $feed_url = $attr["href"];
1510                         }
1511                 }
1512
1513                 return $feed_url;
1514         }
1515
1516         /**
1517          * @brief Check for feed contact
1518          *
1519          * @param string  $url   Profile link
1520          * @param boolean $probe Do a probe if the page contains a feed link
1521          *
1522          * @return array feed data
1523          */
1524         private static function feed($url, $probe = true)
1525         {
1526                 $ret = Network::curl($url);
1527                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1528                         return false;
1529                 }
1530                 $feed = $ret['body'];
1531                 $dummy1 = $dummy2 = $dummy3 = null;
1532                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1533
1534                 if (!$feed_data) {
1535                         if (!$probe) {
1536                                 return false;
1537                         }
1538
1539                         $feed_url = self::getFeedLink($url);
1540
1541                         if (!$feed_url) {
1542                                 return false;
1543                         }
1544
1545                         return self::feed($feed_url, false);
1546                 }
1547
1548                 if (!empty($feed_data["header"]["author-name"])) {
1549                         $data["name"] = $feed_data["header"]["author-name"];
1550                 }
1551
1552                 if (!empty($feed_data["header"]["author-nick"])) {
1553                         $data["nick"] = $feed_data["header"]["author-nick"];
1554                 }
1555
1556                 if (!empty($feed_data["header"]["author-avatar"])) {
1557                         $data["photo"] = $feed_data["header"]["author-avatar"];
1558                 }
1559
1560                 if (!empty($feed_data["header"]["author-id"])) {
1561                         $data["alias"] = $feed_data["header"]["author-id"];
1562                 }
1563
1564                 $data["url"] = $url;
1565                 $data["poll"] = $url;
1566
1567                 if (!empty($feed_data["header"]["author-link"])) {
1568                         $data["baseurl"] = $feed_data["header"]["author-link"];
1569                 } else {
1570                         $data["baseurl"] = $data["url"];
1571                 }
1572
1573                 $data["network"] = NETWORK_FEED;
1574
1575                 return $data;
1576         }
1577
1578         /**
1579          * @brief Check for mail contact
1580          *
1581          * @param string  $uri Profile link
1582          * @param integer $uid User ID
1583          *
1584          * @return array mail data
1585          */
1586         private static function mail($uri, $uid)
1587         {
1588                 if (!Network::isEmailDomainValid($uri)) {
1589                         return false;
1590                 }
1591
1592                 if ($uid == 0) {
1593                         return false;
1594                 }
1595
1596                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1597
1598                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1599
1600                 if (DBM::is_result($x) && DBM::is_result($r)) {
1601                         $mailbox = Email::constructMailboxName($r[0]);
1602                         $password = '';
1603                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1604                         $mbox = Email::connect($mailbox, $r[0]['user'], $password);
1605                         if (!$mbox) {
1606                                 return false;
1607                         }
1608                 }
1609
1610                 $msgs = Email::poll($mbox, $uri);
1611                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1612
1613                 if (!count($msgs)) {
1614                         return false;
1615                 }
1616
1617                 $phost = substr($uri, strpos($uri, '@') + 1);
1618
1619                 $data = [];
1620                 $data["addr"]    = $uri;
1621                 $data["network"] = NETWORK_MAIL;
1622                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1623                 $data["nick"]    = $data["name"];
1624                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1625                 $data["url"]     = 'mailto:'.$uri;
1626                 $data["notify"]  = 'smtp '.random_string();
1627                 $data["poll"]    = 'email '.random_string();
1628
1629                 $x = Email::messageMeta($mbox, $msgs[0]);
1630                 if (stristr($x[0]->from, $uri)) {
1631                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1632                 } elseif (stristr($x[0]->to, $uri)) {
1633                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1634                 }
1635                 if (isset($adr)) {
1636                         foreach ($adr as $feadr) {
1637                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1638                                         &&(strcasecmp($feadr->host, $phost) == 0)
1639                                         && (strlen($feadr->personal))
1640                                 ) {
1641                                         $personal = imap_mime_header_decode($feadr->personal);
1642                                         $data["name"] = "";
1643                                         foreach ($personal as $perspart) {
1644                                                 if ($perspart->charset != "default") {
1645                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1646                                                 } else {
1647                                                         $data["name"] .= $perspart->text;
1648                                                 }
1649                                         }
1650
1651                                         $data["name"] = notags($data["name"]);
1652                                 }
1653                         }
1654                 }
1655                 if (!empty($mbox)) {
1656                         imap_close($mbox);
1657                 }
1658
1659                 return $data;
1660         }
1661
1662         /**
1663          * @brief Mix two paths together to possibly fix missing parts
1664          *
1665          * @param string $avatar Path to the avatar
1666          * @param string $base   Another path that is hopefully complete
1667          *
1668          * @return string fixed avatar path
1669          */
1670         public static function fixAvatar($avatar, $base)
1671         {
1672                 $base_parts = parse_url($base);
1673
1674                 // Remove all parts that could create a problem
1675                 unset($base_parts['path']);
1676                 unset($base_parts['query']);
1677                 unset($base_parts['fragment']);
1678
1679                 $avatar_parts = parse_url($avatar);
1680
1681                 // Now we mix them
1682                 $parts = array_merge($base_parts, $avatar_parts);
1683
1684                 // And put them together again
1685                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1686                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1687                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1688                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1689                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1690                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1691
1692                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1693
1694                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1695
1696                 return $fixed;
1697         }
1698 }