src/Network/Probe.php

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