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 use DomXPath;
  26
  27 require_once 'include/dba.php';
  28
  29 /**
  30  * @brief This class contain functions for probing URL
  31  *
  32  */
  33 class Probe
  34 {
  35         private static $baseurl;
  36
  37         /**
  38          * @brief Rearrange the array so that it always has the same order
  39          *
  40          * @param array $data Unordered data
  41          *
  42          * @return array Ordered data
  43          */
  44         private static function rearrangeData($data)
  45         {
  46                 $fields = ["name", "nick", "guid", "url", "addr", "alias",
  47                                 "photo", "community", "keywords", "location", "about",
  48                                 "batch", "notify", "poll", "request", "confirm", "poco",
  49                                 "priority", "network", "pubkey", "baseurl"];
  50
  51                 $newdata = [];
  52                 foreach ($fields as $field) {
  53                         if (isset($data[$field])) {
  54                                 $newdata[$field] = $data[$field];
  55                         } else {
  56                                 $newdata[$field] = "";
  57                         }
  58                 }
  59
  60                 // We don't use the "priority" field anymore and replace it with a dummy.
  61                 $newdata["priority"] = 0;
  62
  63                 return $newdata;
  64         }
  65
  66         /**
  67          * @brief Check if the hostname belongs to the own server
  68          *
  69          * @param string $host The hostname that is to be checked
  70          *
  71          * @return bool Does the testes hostname belongs to the own server?
  72          */
  73         private static function ownHost($host)
  74         {
  75                 $own_host = get_app()->get_hostname();
  76
  77                 $parts = parse_url($host);
  78
  79                 if (!isset($parts['scheme'])) {
  80                         $parts = parse_url('http://'.$host);
  81                 }
  82
  83                 if (!isset($parts['host'])) {
  84                         return false;
  85                 }
  86                 return $parts['host'] == $own_host;
  87         }
  88
  89         /**
  90          * @brief Probes for webfinger path via "host-meta"
  91          *
  92          * We have to check if the servers in the future still will offer this.
  93          * It seems as if it was dropped from the standard.
  94          *
  95          * @param string $host The host part of an url
  96          *
  97          * @return array with template and type of the webfinger template for JSON or XML
  98          */
  99         private static function hostMeta($host)
 100         {
 101                 // Reset the static variable
 102                 self::$baseurl = '';
 103
 104                 $ssl_url = "https://".$host."/.well-known/host-meta";
 105                 $url = "http://".$host."/.well-known/host-meta";
 106
 107                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 108                 $redirects = 0;
 109
 110                 logger("Probing for ".$host, LOGGER_DEBUG);
 111                 $xrd = null;
 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["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 123                                 logger("Probing timeout for ".$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                         $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($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($data['nick'], ' ')));
 357                         }
 358                 }
 359
 360                 if (!empty(self::$baseurl)) {
 361                         $data["baseurl"] = self::$baseurl;
 362                 }
 363
 364                 if (empty($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' => $data['baseurl']];
 397
 398                                 // This doesn't cover the case when a community isn't a community anymore
 399                                 if (!empty($data['community']) && $data['community']) {
 400                                         $fields['community'] = $data['community'];
 401                                         $fields['contact-type'] = ACCOUNT_TYPE_COMMUNITY;
 402                                 }
 403
 404                                 $fieldnames = [];
 405
 406                                 foreach ($fields as $key => $val) {
 407                                         if (empty($val)) {
 408                                                 unset($fields[$key]);
 409                                         } else {
 410                                                 $fieldnames[] = $key;
 411                                         }
 412                                 }
 413
 414                                 $fields['updated'] = DBM::date();
 415
 416                                 $condition = ['nurl' => normalise_link($data["url"])];
 417
 418                                 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
 419
 420                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 421                                 // In difference to the public contacts we want to have every contact
 422                                 // in the world in our global contacts.
 423                                 if (!$old_fields) {
 424                                         $old_fields = true;
 425
 426                                         // These values have to be set only on insert
 427                                         $fields['photo'] = $data['photo'];
 428                                         $fields['created'] = DateTimeFormat::utcNow();
 429                                 }
 430
 431                                 DBA::update('gcontact', $fields, $condition, $old_fields);
 432
 433                                 $fields = ['name' => $data['name'],
 434                                                 'nick' => $data['nick'],
 435                                                 'url' => $data['url'],
 436                                                 'addr' => $data['addr'],
 437                                                 'alias' => $data['alias'],
 438                                                 'keywords' => $data['keywords'],
 439                                                 'location' => $data['location'],
 440                                                 'about' => $data['about'],
 441                                                 'batch' => $data['batch'],
 442                                                 'notify' => $data['notify'],
 443                                                 'poll' => $data['poll'],
 444                                                 'request' => $data['request'],
 445                                                 'confirm' => $data['confirm'],
 446                                                 'poco' => $data['poco'],
 447                                                 'network' => $data['network'],
 448                                                 'pubkey' => $data['pubkey'],
 449                                                 'priority' => $data['priority'],
 450                                                 'writable' => true,
 451                                                 'rel' => CONTACT_IS_SHARING];
 452
 453                                 $fieldnames = [];
 454
 455                                 foreach ($fields as $key => $val) {
 456                                         if (empty($val)) {
 457                                                 unset($fields[$key]);
 458                                         } else {
 459                                                 $fieldnames[] = $key;
 460                                         }
 461                                 }
 462
 463                                 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 464
 465                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 466                                 // This won't trigger an insert. This is intended, since we only need
 467                                 // public contacts for everyone we store items from.
 468                                 // We don't need to store every contact on the planet.
 469                                 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 470
 471                                 $fields['name-date'] = DateTimeFormat::utcNow();
 472                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 473                                 $fields['success_update'] = DateTimeFormat::utcNow();
 474
 475                                 DBA::update('contact', $fields, $condition, $old_fields);
 476                         }
 477                 }
 478
 479                 return $data;
 480         }
 481
 482         /**
 483          * @brief Switch the scheme of an url between http and https
 484          *
 485          * @param string $url URL
 486          *
 487          * @return string switched URL
 488          */
 489         private static function switchScheme($url)
 490         {
 491                 $parts = parse_url($url);
 492
 493                 if (!isset($parts['scheme'])) {
 494                         return $url;
 495                 }
 496
 497                 if ($parts['scheme'] == 'http') {
 498                         $url = str_replace('http://', 'https://', $url);
 499                 } elseif ($parts['scheme'] == 'https') {
 500                         $url = str_replace('https://', 'http://', $url);
 501                 }
 502
 503                 return $url;
 504         }
 505
 506         /**
 507          * @brief Checks if a profile url should be OStatus but only provides partial information
 508          *
 509          * @param array  $webfinger Webfinger data
 510          * @param string $lrdd      Path template for webfinger request
 511          * @param string $type      type
 512          *
 513          * @return array fixed webfinger data
 514          */
 515         private static function fixOStatus($webfinger, $lrdd, $type)
 516         {
 517                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 518                         return $webfinger;
 519                 }
 520
 521                 $is_ostatus = false;
 522                 $has_key = false;
 523
 524                 foreach ($webfinger['links'] as $link) {
 525                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 526                                 $is_ostatus = true;
 527                         }
 528                         if ($link['rel'] == 'magic-public-key') {
 529                                 $has_key = true;
 530                         }
 531                 }
 532
 533                 if (!$is_ostatus || $has_key) {
 534                         return $webfinger;
 535                 }
 536
 537                 $url = self::switchScheme($webfinger['subject']);
 538                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 539                 $webfinger2 = self::webfinger($path, $type);
 540
 541                 // Is the new webfinger detectable as OStatus?
 542                 if (self::ostatus($webfinger2, true)) {
 543                         $webfinger = $webfinger2;
 544                 }
 545
 546                 return $webfinger;
 547         }
 548
 549         /**
 550          * @brief Fetch information (protocol endpoints and user information) about a given uri
 551          *
 552          * This function is only called by the "uri" function that adds caching and rearranging of data.
 553          *
 554          * @param string  $uri     Address that should be probed
 555          * @param string  $network Test for this specific network
 556          * @param integer $uid     User ID for the probe (only used for mails)
 557          *
 558          * @return array uri data
 559          */
 560         private static function detect($uri, $network, $uid)
 561         {
 562                 $parts = parse_url($uri);
 563
 564                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 565                         $host = $parts["host"];
 566                         if (!empty($parts["port"])) {
 567                                 $host .= ':'.$parts["port"];
 568                         }
 569
 570                         if ($host == 'twitter.com') {
 571                                 return ["network" => NETWORK_TWITTER];
 572                         }
 573                         $lrdd = self::hostMeta($host);
 574
 575                         if (is_bool($lrdd)) {
 576                                 return [];
 577                         }
 578
 579                         $path_parts = explode("/", trim($parts["path"], "/"));
 580
 581                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 582                                 $host .= "/".array_shift($path_parts);
 583                                 $lrdd = self::hostMeta($host);
 584                         }
 585                         if (!$lrdd) {
 586                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 587                                 return self::feed($uri);
 588                         }
 589                         $nick = array_pop($path_parts);
 590
 591                         // Mastodon uses a "@" as prefix for usernames in their url format
 592                         $nick = ltrim($nick, '@');
 593
 594                         $addr = $nick."@".$host;
 595                 } elseif (strstr($uri, '@')) {
 596                         // If the URI starts with "mailto:" then jump directly to the mail detection
 597                         if (strpos($uri, 'mailto:') !== false) {
 598                                 $uri = str_replace('mailto:', '', $uri);
 599                                 return self::mail($uri, $uid);
 600                         }
 601
 602                         if ($network == NETWORK_MAIL) {
 603                                 return self::mail($uri, $uid);
 604                         }
 605                         // Remove "acct:" from the URI
 606                         $uri = str_replace('acct:', '', $uri);
 607
 608                         $host = substr($uri, strpos($uri, '@') + 1);
 609                         $nick = substr($uri, 0, strpos($uri, '@'));
 610
 611                         if (strpos($uri, '@twitter.com')) {
 612                                 return ["network" => NETWORK_TWITTER];
 613                         }
 614                         $lrdd = self::hostMeta($host);
 615
 616                         if (is_bool($lrdd)) {
 617                                 return [];
 618                         }
 619
 620                         if (!$lrdd) {
 621                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 622                                 return self::mail($uri, $uid);
 623                         }
 624                         $addr = $uri;
 625                 } else {
 626                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 627                         return false;
 628                 }
 629
 630                 $webfinger = false;
 631
 632                 /// @todo Do we need the prefix "acct:" or "acct://"?
 633
 634                 foreach ($lrdd as $type => $template) {
 635                         if ($webfinger) {
 636                                 continue;
 637                         }
 638
 639                         // At first try it with the given uri
 640                         $path = str_replace('{uri}', urlencode($uri), $template);
 641                         $webfinger = self::webfinger($path, $type);
 642
 643                         // Fix possible problems with GNU Social probing to wrong scheme
 644                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 645
 646                         // We cannot be sure that the detected address was correct, so we don't use the values
 647                         if ($webfinger && ($uri != $addr)) {
 648                                 $nick = "";
 649                                 $addr = "";
 650                         }
 651
 652                         // Try webfinger with the address (user@domain.tld)
 653                         if (!$webfinger) {
 654                                 $path = str_replace('{uri}', urlencode($addr), $template);
 655                                 $webfinger = self::webfinger($path, $type);
 656                         }
 657
 658                         // Mastodon needs to have it with "acct:"
 659                         if (!$webfinger) {
 660                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 661                                 $webfinger = self::webfinger($path, $type);
 662                         }
 663                 }
 664
 665                 if (!$webfinger) {
 666                         return self::feed($uri);
 667                 }
 668
 669                 $result = false;
 670
 671                 logger("Probing ".$uri, LOGGER_DEBUG);
 672
 673                 if (in_array($network, ["", NETWORK_DFRN])) {
 674                         $result = self::dfrn($webfinger);
 675                 }
 676                 if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
 677                         $result = self::diaspora($webfinger);
 678                 }
 679                 if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
 680                         $result = self::ostatus($webfinger);
 681                 }
 682                 if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
 683                         $result = self::pumpio($webfinger, $addr);
 684                 }
 685                 if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
 686                         $result = self::feed($uri);
 687                 } else {
 688                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 689                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 690                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 691                                 $result["nick"] = $nick;
 692                         }
 693
 694                         if (empty($result["addr"]) && ($addr != "")) {
 695                                 $result["addr"] = $addr;
 696                         }
 697                 }
 698
 699                 if (empty($result["network"])) {
 700                         $result["network"] = NETWORK_PHANTOM;
 701                 }
 702
 703                 if (empty($result["url"])) {
 704                         $result["url"] = $uri;
 705                 }
 706
 707                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 708
 709                 if (empty($result["baseurl"])) {
 710                         $pos = strpos($result["url"], $host);
 711                         if ($pos) {
 712                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 713                         }
 714                 }
 715                 return $result;
 716         }
 717
 718         /**
 719          * @brief Perform a webfinger request.
 720          *
 721          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 722          *
 723          * @param string $url  Address that should be probed
 724          * @param string $type type
 725          *
 726          * @return array webfinger data
 727          */
 728         private static function webfinger($url, $type)
 729         {
 730                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 731                 $redirects = 0;
 732
 733                 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 734                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 735                         return false;
 736                 }
 737                 $data = $ret['body'];
 738
 739                 $webfinger = json_decode($data, true);
 740                 if (is_array($webfinger)) {
 741                         if (!isset($webfinger["links"])) {
 742                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 743                                 return false;
 744                         }
 745                         return $webfinger;
 746                 }
 747
 748                 // If it is not JSON, maybe it is XML
 749                 $xrd = XML::parseString($data, false);
 750                 if (!is_object($xrd)) {
 751                         logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
 752                         return false;
 753                 }
 754
 755                 $xrd_arr = XML::elementToArray($xrd);
 756                 if (!isset($xrd_arr["xrd"]["link"])) {
 757                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 758                         return false;
 759                 }
 760
 761                 $webfinger = [];
 762
 763                 if (!empty($xrd_arr["xrd"]["subject"])) {
 764                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 765                 }
 766
 767                 if (!empty($xrd_arr["xrd"]["alias"])) {
 768                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 769                 }
 770
 771                 $webfinger["links"] = [];
 772
 773                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 774                         if (!empty($data["@attributes"])) {
 775                                 $attributes = $data["@attributes"];
 776                         } elseif ($value == "@attributes") {
 777                                 $attributes = $data;
 778                         } else {
 779                                 continue;
 780                         }
 781
 782                         $webfinger["links"][] = $attributes;
 783                 }
 784                 return $webfinger;
 785         }
 786
 787         /**
 788          * @brief Poll the Friendica specific noscrape page.
 789          *
 790          * "noscrape" is a faster alternative to fetch the data from the hcard.
 791          * This functionality was originally created for the directory.
 792          *
 793          * @param string $noscrape_url Link to the noscrape page
 794          * @param array  $data         The already fetched data
 795          *
 796          * @return array noscrape data
 797          */
 798         private static function pollNoscrape($noscrape_url, $data)
 799         {
 800                 $ret = Network::curl($noscrape_url);
 801                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 802                         return false;
 803                 }
 804                 $content = $ret['body'];
 805                 if (!$content) {
 806                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 807                         return false;
 808                 }
 809
 810                 $json = json_decode($content, true);
 811                 if (!is_array($json)) {
 812                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 813                         return false;
 814                 }
 815
 816                 if (!empty($json["fn"])) {
 817                         $data["name"] = $json["fn"];
 818                 }
 819
 820                 if (!empty($json["addr"])) {
 821                         $data["addr"] = $json["addr"];
 822                 }
 823
 824                 if (!empty($json["nick"])) {
 825                         $data["nick"] = $json["nick"];
 826                 }
 827
 828                 if (!empty($json["guid"])) {
 829                         $data["guid"] = $json["guid"];
 830                 }
 831
 832                 if (!empty($json["comm"])) {
 833                         $data["community"] = $json["comm"];
 834                 }
 835
 836                 if (!empty($json["tags"])) {
 837                         $keywords = implode(" ", $json["tags"]);
 838                         if ($keywords != "") {
 839                                 $data["keywords"] = $keywords;
 840                         }
 841                 }
 842
 843                 $location = Profile::formatLocation($json);
 844                 if ($location) {
 845                         $data["location"] = $location;
 846                 }
 847
 848                 if (!empty($json["about"])) {
 849                         $data["about"] = $json["about"];
 850                 }
 851
 852                 if (!empty($json["key"])) {
 853                         $data["pubkey"] = $json["key"];
 854                 }
 855
 856                 if (!empty($json["photo"])) {
 857                         $data["photo"] = $json["photo"];
 858                 }
 859
 860                 if (!empty($json["dfrn-request"])) {
 861                         $data["request"] = $json["dfrn-request"];
 862                 }
 863
 864                 if (!empty($json["dfrn-confirm"])) {
 865                         $data["confirm"] = $json["dfrn-confirm"];
 866                 }
 867
 868                 if (!empty($json["dfrn-notify"])) {
 869                         $data["notify"] = $json["dfrn-notify"];
 870                 }
 871
 872                 if (!empty($json["dfrn-poll"])) {
 873                         $data["poll"] = $json["dfrn-poll"];
 874                 }
 875
 876                 return $data;
 877         }
 878
 879         /**
 880          * @brief Check for valid DFRN data
 881          *
 882          * @param array $data DFRN data
 883          *
 884          * @return int Number of errors
 885          */
 886         public static function validDfrn($data)
 887         {
 888                 $errors = 0;
 889                 if (!isset($data['key'])) {
 890                         $errors ++;
 891                 }
 892                 if (!isset($data['dfrn-request'])) {
 893                         $errors ++;
 894                 }
 895                 if (!isset($data['dfrn-confirm'])) {
 896                         $errors ++;
 897                 }
 898                 if (!isset($data['dfrn-notify'])) {
 899                         $errors ++;
 900                 }
 901                 if (!isset($data['dfrn-poll'])) {
 902                         $errors ++;
 903                 }
 904                 return $errors;
 905         }
 906
 907         /**
 908          * @brief Fetch data from a DFRN profile page and via "noscrape"
 909          *
 910          * @param string $profile_link Link to the profile page
 911          *
 912          * @return array profile data
 913          */
 914         public static function profile($profile_link)
 915         {
 916                 $data = [];
 917
 918                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 919
 920                 // Fetch data via noscrape - this is faster
 921                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 922                 $data = self::pollNoscrape($noscrape_url, $data);
 923
 924                 if (!isset($data["notify"])
 925                         || !isset($data["confirm"])
 926                         || !isset($data["request"])
 927                         || !isset($data["poll"])
 928                         || !isset($data["poco"])
 929                         || !isset($data["name"])
 930                         || !isset($data["photo"])
 931                 ) {
 932                         $data = self::pollHcard($profile_link, $data, true);
 933                 }
 934
 935                 $prof_data = [];
 936                 $prof_data["addr"]         = $data["addr"];
 937                 $prof_data["nick"]         = $data["nick"];
 938                 $prof_data["dfrn-request"] = $data["request"];
 939                 $prof_data["dfrn-confirm"] = $data["confirm"];
 940                 $prof_data["dfrn-notify"]  = $data["notify"];
 941                 $prof_data["dfrn-poll"]    = $data["poll"];
 942                 $prof_data["dfrn-poco"]    = $data["poco"];
 943                 $prof_data["photo"]        = $data["photo"];
 944                 $prof_data["fn"]           = $data["name"];
 945                 $prof_data["key"]          = $data["pubkey"];
 946
 947                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 948
 949                 return $prof_data;
 950         }
 951
 952         /**
 953          * @brief Check for DFRN contact
 954          *
 955          * @param array $webfinger Webfinger data
 956          *
 957          * @return array DFRN data
 958          */
 959         private static function dfrn($webfinger)
 960         {
 961                 $hcard_url = "";
 962                 $data = [];
 963                 foreach ($webfinger["links"] as $link) {
 964                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 965                                 $data["network"] = NETWORK_DFRN;
 966                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 967                                 $data["poll"] = $link["href"];
 968                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 969                                 $data["url"] = $link["href"];
 970                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 971                                 $hcard_url = $link["href"];
 972                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 973                                 $data["poco"] = $link["href"];
 974                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 975                                 $data["photo"] = $link["href"];
 976                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 977                                 $data["baseurl"] = trim($link["href"], '/');
 978                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 979                                 $data["guid"] = $link["href"];
 980                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 981                                 $data["pubkey"] = base64_decode($link["href"]);
 982
 983                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 984                                 if (strstr($data["pubkey"], 'RSA ')) {
 985                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
 986                                 }
 987                         }
 988                 }
 989
 990                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 991                         foreach ($webfinger["aliases"] as $alias) {
 992                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
 993                                         $data["alias"] = $alias;
 994                                 } elseif (substr($alias, 0, 5) == 'acct:') {
 995                                         $data["addr"] = substr($alias, 5);
 996                                 }
 997                         }
 998                 }
 999
1000                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1001                         $data["addr"] = substr($webfinger["subject"], 5);
1002                 }
1003
1004                 if (!isset($data["network"]) || ($hcard_url == "")) {
1005                         return false;
1006                 }
1007
1008                 // Fetch data via noscrape - this is faster
1009                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1010                 $data = self::pollNoscrape($noscrape_url, $data);
1011
1012                 if (isset($data["notify"])
1013                         && isset($data["confirm"])
1014                         && isset($data["request"])
1015                         && isset($data["poll"])
1016                         && isset($data["name"])
1017                         && isset($data["photo"])
1018                 ) {
1019                         return $data;
1020                 }
1021
1022                 $data = self::pollHcard($hcard_url, $data, true);
1023
1024                 return $data;
1025         }
1026
1027         /**
1028          * @brief Poll the hcard page (Diaspora and Friendica specific)
1029          *
1030          * @param string  $hcard_url Link to the hcard page
1031          * @param array   $data      The already fetched data
1032          * @param boolean $dfrn      Poll DFRN specific data
1033          *
1034          * @return array hcard data
1035          */
1036         private static function pollHcard($hcard_url, $data, $dfrn = false)
1037         {
1038                 $ret = Network::curl($hcard_url);
1039                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1040                         return false;
1041                 }
1042                 $content = $ret['body'];
1043                 if (!$content) {
1044                         return false;
1045                 }
1046
1047                 $doc = new DOMDocument();
1048                 if (!@$doc->loadHTML($content)) {
1049                         return false;
1050                 }
1051
1052                 $xpath = new DomXPath($doc);
1053
1054                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1055                 if (!is_object($vcards)) {
1056                         return false;
1057                 }
1058
1059                 if (!isset($data["baseurl"])) {
1060                         $data["baseurl"] = "";
1061                 }
1062
1063                 if ($vcards->length > 0) {
1064                         $vcard = $vcards->item(0);
1065
1066                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1067                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1068                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1069                         if (($search->length > 0) && empty($data["guid"])) {
1070                                 $data["guid"] = $search->item(0)->nodeValue;
1071                         }
1072
1073                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1074                         if ($search->length > 0) {
1075                                 $data["nick"] = $search->item(0)->nodeValue;
1076                         }
1077
1078                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1079                         if ($search->length > 0) {
1080                                 $data["name"] = $search->item(0)->nodeValue;
1081                         }
1082
1083                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1084                         if ($search->length > 0) {
1085                                 $data["searchable"] = $search->item(0)->nodeValue;
1086                         }
1087
1088                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1089                         if ($search->length > 0) {
1090                                 $data["pubkey"] = $search->item(0)->nodeValue;
1091                                 if (strstr($data["pubkey"], 'RSA ')) {
1092                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1093                                 }
1094                         }
1095
1096                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1097                         if ($search->length > 0) {
1098                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1099                         }
1100                 }
1101
1102                 $avatar = [];
1103                 if (!empty($vcard)) {
1104                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1105                         foreach ($photos as $photo) {
1106                                 $attr = [];
1107                                 foreach ($photo->attributes as $attribute) {
1108                                         $attr[$attribute->name] = trim($attribute->value);
1109                                 }
1110
1111                                 if (isset($attr["src"]) && isset($attr["width"])) {
1112                                         $avatar[$attr["width"]] = $attr["src"];
1113                                 }
1114
1115                                 // We don't have a width. So we just take everything that we got.
1116                                 // This is a Hubzilla workaround which doesn't send a width.
1117                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1118                                         $avatar[] = $attr["src"];
1119                                 }
1120                         }
1121                 }
1122
1123                 if (sizeof($avatar)) {
1124                         ksort($avatar);
1125                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1126                 }
1127
1128                 if ($dfrn) {
1129                         // Poll DFRN specific data
1130                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1131                         if ($search->length > 0) {
1132                                 foreach ($search as $link) {
1133                                         //$data["request"] = $search->item(0)->nodeValue;
1134                                         $attr = [];
1135                                         foreach ($link->attributes as $attribute) {
1136                                                 $attr[$attribute->name] = trim($attribute->value);
1137                                         }
1138
1139                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1140                                 }
1141                         }
1142
1143                         // Older Friendica versions had used the "uid" field differently than newer versions
1144                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1145                                 unset($data["guid"]);
1146                         }
1147                 }
1148
1149
1150                 return $data;
1151         }
1152
1153         /**
1154          * @brief Check for Diaspora contact
1155          *
1156          * @param array $webfinger Webfinger data
1157          *
1158          * @return array Diaspora data
1159          */
1160         private static function diaspora($webfinger)
1161         {
1162                 $hcard_url = "";
1163                 $data = [];
1164                 foreach ($webfinger["links"] as $link) {
1165                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1166                                 $hcard_url = $link["href"];
1167                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1168                                 $data["baseurl"] = trim($link["href"], '/');
1169                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1170                                 $data["guid"] = $link["href"];
1171                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1172                                 $data["url"] = $link["href"];
1173                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1174                                 $data["poll"] = $link["href"];
1175                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1176                                 $data["poco"] = $link["href"];
1177                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1178                                 $data["notify"] = $link["href"];
1179                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1180                                 $data["pubkey"] = base64_decode($link["href"]);
1181
1182                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1183                                 if (strstr($data["pubkey"], 'RSA ')) {
1184                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1185                                 }
1186                         }
1187                 }
1188
1189                 if (!isset($data["url"]) || ($hcard_url == "")) {
1190                         return false;
1191                 }
1192
1193                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1194                         foreach ($webfinger["aliases"] as $alias) {
1195                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1196                                         $data["alias"] = $alias;
1197                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1198                                         $data["addr"] = substr($alias, 5);
1199                                 }
1200                         }
1201                 }
1202
1203                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1204                         $data["addr"] = substr($webfinger["subject"], 5);
1205                 }
1206
1207                 // Fetch further information from the hcard
1208                 $data = self::pollHcard($hcard_url, $data);
1209
1210                 if (!$data) {
1211                         return false;
1212                 }
1213
1214                 if (isset($data["url"])
1215                         && isset($data["guid"])
1216                         && isset($data["baseurl"])
1217                         && isset($data["pubkey"])
1218                         && ($hcard_url != "")
1219                 ) {
1220                         $data["network"] = NETWORK_DIASPORA;
1221
1222                         // The Diaspora handle must always be lowercase
1223                         if (!empty($data["addr"])) {
1224                                 $data["addr"] = strtolower($data["addr"]);
1225                         }
1226
1227                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1228                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1229                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1230                 } else {
1231                         return false;
1232                 }
1233
1234                 return $data;
1235         }
1236
1237         /**
1238          * @brief Check for OStatus contact
1239          *
1240          * @param array $webfinger Webfinger data
1241          * @param bool  $short     Short detection mode
1242          *
1243          * @return array|bool OStatus data or "false" on error or "true" on short mode
1244          */
1245         private static function ostatus($webfinger, $short = false)
1246         {
1247                 $data = [];
1248
1249                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1250                         foreach ($webfinger["aliases"] as $alias) {
1251                                 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1252                                         $data["addr"] = str_replace('acct:', '', $alias);
1253                                 }
1254                         }
1255                 }
1256
1257                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1258                         && !strstr(normalise_link($webfinger["subject"]), "http://")
1259                 ) {
1260                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1261                 }
1262
1263                 $pubkey = "";
1264                 if (is_array($webfinger["links"])) {
1265                         foreach ($webfinger["links"] as $link) {
1266                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1267                                         && ($link["type"] == "text/html")
1268                                         && ($link["href"] != "")
1269                                 ) {
1270                                         $data["url"] = $link["href"];
1271                                 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1272                                         $data["notify"] = $link["href"];
1273                                 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1274                                         $data["poll"] = $link["href"];
1275                                 } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1276                                         $pubkey = $link["href"];
1277
1278                                         if (substr($pubkey, 0, 5) === 'data:') {
1279                                                 if (strstr($pubkey, ',')) {
1280                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1281                                                 } else {
1282                                                         $pubkey = substr($pubkey, 5);
1283                                                 }
1284                                         } elseif (normalise_link($pubkey) == 'http://') {
1285                                                 $ret = Network::curl($pubkey);
1286                                                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1287                                                         return false;
1288                                                 }
1289                                                 $pubkey = $ret['body'];
1290                                         }
1291
1292                                         $key = explode(".", $pubkey);
1293
1294                                         if (sizeof($key) >= 3) {
1295                                                 $m = base64url_decode($key[1]);
1296                                                 $e = base64url_decode($key[2]);
1297                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1298                                         }
1299                                 }
1300                         }
1301                 }
1302
1303                 if (isset($data["notify"]) && isset($data["pubkey"])
1304                         && isset($data["poll"])
1305                         && isset($data["url"])
1306                 ) {
1307                         $data["network"] = NETWORK_OSTATUS;
1308                 } else {
1309                         return false;
1310                 }
1311
1312                 if ($short) {
1313                         return true;
1314                 }
1315
1316                 // Fetch all additional data from the feed
1317                 $ret = Network::curl($data["poll"]);
1318                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1319                         return false;
1320                 }
1321                 $feed = $ret['body'];
1322                 $dummy1 = null;
1323                 $dummy2 = null;
1324                 $dummy2 = null;
1325                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1326                 if (!$feed_data) {
1327                         return false;
1328                 }
1329
1330                 if (!empty($feed_data["header"]["author-name"])) {
1331                         $data["name"] = $feed_data["header"]["author-name"];
1332                 }
1333                 if (!empty($feed_data["header"]["author-nick"])) {
1334                         $data["nick"] = $feed_data["header"]["author-nick"];
1335                 }
1336                 if (!empty($feed_data["header"]["author-avatar"])) {
1337                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1338                 }
1339                 if (!empty($feed_data["header"]["author-id"])) {
1340                         $data["alias"] = $feed_data["header"]["author-id"];
1341                 }
1342                 if (!empty($feed_data["header"]["author-location"])) {
1343                         $data["location"] = $feed_data["header"]["author-location"];
1344                 }
1345                 if (!empty($feed_data["header"]["author-about"])) {
1346                         $data["about"] = $feed_data["header"]["author-about"];
1347                 }
1348                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1349                 // So we take the value that we just fetched, although the other one worked as well
1350                 if (!empty($feed_data["header"]["author-link"])) {
1351                         $data["url"] = $feed_data["header"]["author-link"];
1352                 }
1353
1354                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1355                         $data['url'] = $data["alias"];
1356                         $data["alias"] = '';
1357                 }
1358
1359                 /// @todo Fetch location and "about" from the feed as well
1360                 return $data;
1361         }
1362
1363         /**
1364          * @brief Fetch data from a pump.io profile page
1365          *
1366          * @param string $profile_link Link to the profile page
1367          *
1368          * @return array profile data
1369          */
1370         private static function pumpioProfileData($profile_link)
1371         {
1372                 $doc = new DOMDocument();
1373                 if (!@$doc->loadHTMLFile($profile_link)) {
1374                         return false;
1375                 }
1376
1377                 $xpath = new DomXPath($doc);
1378
1379                 $data = [];
1380
1381                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1382
1383                 if ($data["name"] == '') {
1384                         // This is ugly - but pump.io doesn't seem to know a better way for it
1385                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1386                         $pos = strpos($data["name"], chr(10));
1387                         if ($pos) {
1388                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1389                         }
1390                 }
1391
1392                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1393
1394                 if ($data["location"] == '') {
1395                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1396                 }
1397
1398                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1399
1400                 if ($data["about"] == '') {
1401                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1402                 }
1403
1404                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1405                 if (!$avatar) {
1406                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1407                 }
1408                 if ($avatar) {
1409                         foreach ($avatar->attributes as $attribute) {
1410                                 if ($attribute->name == "src") {
1411                                         $data["photo"] = trim($attribute->value);
1412                                 }
1413                         }
1414                 }
1415
1416                 return $data;
1417         }
1418
1419         /**
1420          * @brief Check for pump.io contact
1421          *
1422          * @param array $webfinger Webfinger data
1423          *
1424          * @return array pump.io data
1425          */
1426         private static function pumpio($webfinger, $addr)
1427         {
1428                 $data = [];
1429                 foreach ($webfinger["links"] as $link) {
1430                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1431                                 && ($link["type"] == "text/html")
1432                                 && ($link["href"] != "")
1433                         ) {
1434                                 $data["url"] = $link["href"];
1435                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1436                                 $data["notify"] = $link["href"];
1437                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1438                                 $data["poll"] = $link["href"];
1439                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1440                                 $data["dialback"] = $link["href"];
1441                         }
1442                 }
1443                 if (isset($data["poll"]) && isset($data["notify"])
1444                         && isset($data["dialback"])
1445                         && isset($data["url"])
1446                 ) {
1447                         // by now we use these fields only for the network type detection
1448                         // So we unset all data that isn't used at the moment
1449                         unset($data["dialback"]);
1450
1451                         $data["network"] = NETWORK_PUMPIO;
1452                 } else {
1453                         return false;
1454                 }
1455
1456                 $profile_data = self::pumpioProfileData($data["url"]);
1457
1458                 if (!$profile_data) {
1459                         return false;
1460                 }
1461
1462                 $data = array_merge($data, $profile_data);
1463
1464                 if (($addr != '') && ($data['name'] != '')) {
1465                         $name = trim(str_replace($addr, '', $data['name']));
1466                         if ($name != '') {
1467                                 $data['name'] = $name;
1468                         }
1469                 }
1470
1471                 return $data;
1472         }
1473
1474         /**
1475          * @brief Check page for feed link
1476          *
1477          * @param string $url Page link
1478          *
1479          * @return string feed link
1480          */
1481         private static function getFeedLink($url)
1482         {
1483                 $doc = new DOMDocument();
1484
1485                 if (!@$doc->loadHTMLFile($url)) {
1486                         return false;
1487                 }
1488
1489                 $xpath = new DomXPath($doc);
1490
1491                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1492                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1493                 if (!is_object($feeds)) {
1494                         return false;
1495                 }
1496
1497                 if ($feeds->length == 0) {
1498                         return false;
1499                 }
1500
1501                 $feed_url = "";
1502
1503                 foreach ($feeds as $feed) {
1504                         $attr = [];
1505                         foreach ($feed->attributes as $attribute) {
1506                                 $attr[$attribute->name] = trim($attribute->value);
1507                         }
1508
1509                         if ($feed_url == "") {
1510                                 $feed_url = $attr["href"];
1511                         }
1512                 }
1513
1514                 return $feed_url;
1515         }
1516
1517         /**
1518          * @brief Check for feed contact
1519          *
1520          * @param string  $url   Profile link
1521          * @param boolean $probe Do a probe if the page contains a feed link
1522          *
1523          * @return array feed data
1524          */
1525         private static function feed($url, $probe = true)
1526         {
1527                 $ret = Network::curl($url);
1528                 if (!$ret["success"] && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1529                         return false;
1530                 }
1531                 $feed = $ret['body'];
1532                 $dummy1 = $dummy2 = $dummy3 = null;
1533                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1534
1535                 if (!$feed_data) {
1536                         if (!$probe) {
1537                                 return false;
1538                         }
1539
1540                         $feed_url = self::getFeedLink($url);
1541
1542                         if (!$feed_url) {
1543                                 return false;
1544                         }
1545
1546                         return self::feed($feed_url, false);
1547                 }
1548
1549                 if (!empty($feed_data["header"]["author-name"])) {
1550                         $data["name"] = $feed_data["header"]["author-name"];
1551                 }
1552
1553                 if (!empty($feed_data["header"]["author-nick"])) {
1554                         $data["nick"] = $feed_data["header"]["author-nick"];
1555                 }
1556
1557                 if (!empty($feed_data["header"]["author-avatar"])) {
1558                         $data["photo"] = $feed_data["header"]["author-avatar"];
1559                 }
1560
1561                 if (!empty($feed_data["header"]["author-id"])) {
1562                         $data["alias"] = $feed_data["header"]["author-id"];
1563                 }
1564
1565                 $data["url"] = $url;
1566                 $data["poll"] = $url;
1567
1568                 if (!empty($feed_data["header"]["author-link"])) {
1569                         $data["baseurl"] = $feed_data["header"]["author-link"];
1570                 } else {
1571                         $data["baseurl"] = $data["url"];
1572                 }
1573
1574                 $data["network"] = NETWORK_FEED;
1575
1576                 return $data;
1577         }
1578
1579         /**
1580          * @brief Check for mail contact
1581          *
1582          * @param string  $uri Profile link
1583          * @param integer $uid User ID
1584          *
1585          * @return array mail data
1586          */
1587         private static function mail($uri, $uid)
1588         {
1589                 if (!Network::isEmailDomainValid($uri)) {
1590                         return false;
1591                 }
1592
1593                 if ($uid == 0) {
1594                         return false;
1595                 }
1596
1597                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1598
1599                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1600
1601                 if (DBM::is_result($x) && DBM::is_result($r)) {
1602                         $mailbox = Email::constructMailboxName($r[0]);
1603                         $password = '';
1604                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1605                         $mbox = Email::connect($mailbox, $r[0]['user'], $password);
1606                         if (!$mbox) {
1607                                 return false;
1608                         }
1609                 }
1610
1611                 $msgs = Email::poll($mbox, $uri);
1612                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1613
1614                 if (!count($msgs)) {
1615                         return false;
1616                 }
1617
1618                 $phost = substr($uri, strpos($uri, '@') + 1);
1619
1620                 $data = [];
1621                 $data["addr"]    = $uri;
1622                 $data["network"] = NETWORK_MAIL;
1623                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1624                 $data["nick"]    = $data["name"];
1625                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1626                 $data["url"]     = 'mailto:'.$uri;
1627                 $data["notify"]  = 'smtp '.random_string();
1628                 $data["poll"]    = 'email '.random_string();
1629
1630                 $x = Email::messageMeta($mbox, $msgs[0]);
1631                 if (stristr($x[0]->from, $uri)) {
1632                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1633                 } elseif (stristr($x[0]->to, $uri)) {
1634                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1635                 }
1636                 if (isset($adr)) {
1637                         foreach ($adr as $feadr) {
1638                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1639                                         &&(strcasecmp($feadr->host, $phost) == 0)
1640                                         && (strlen($feadr->personal))
1641                                 ) {
1642                                         $personal = imap_mime_header_decode($feadr->personal);
1643                                         $data["name"] = "";
1644                                         foreach ($personal as $perspart) {
1645                                                 if ($perspart->charset != "default") {
1646                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1647                                                 } else {
1648                                                         $data["name"] .= $perspart->text;
1649                                                 }
1650                                         }
1651
1652                                         $data["name"] = notags($data["name"]);
1653                                 }
1654                         }
1655                 }
1656                 if (!empty($mbox)) {
1657                         imap_close($mbox);
1658                 }
1659
1660                 return $data;
1661         }
1662
1663         /**
1664          * @brief Mix two paths together to possibly fix missing parts
1665          *
1666          * @param string $avatar Path to the avatar
1667          * @param string $base   Another path that is hopefully complete
1668          *
1669          * @return string fixed avatar path
1670          */
1671         public static function fixAvatar($avatar, $base)
1672         {
1673                 $base_parts = parse_url($base);
1674
1675                 // Remove all parts that could create a problem
1676                 unset($base_parts['path']);
1677                 unset($base_parts['query']);
1678                 unset($base_parts['fragment']);
1679
1680                 $avatar_parts = parse_url($avatar);
1681
1682                 // Now we mix them
1683                 $parts = array_merge($base_parts, $avatar_parts);
1684
1685                 // And put them together again
1686                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1687                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1688                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1689                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1690                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1691                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1692
1693                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1694
1695                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1696
1697                 return $fixed;
1698         }
1699 }