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\Protocol;
  16 use Friendica\Core\System;
  17 use Friendica\Database\DBA;
  18 use Friendica\Model\Contact;
  19 use Friendica\Model\Profile;
  20 use Friendica\Protocol\Email;
  21 use Friendica\Protocol\Feed;
  22 use Friendica\Protocol\ActivityPub;
  23 use Friendica\Util\Crypto;
  24 use Friendica\Util\DateTimeFormat;
  25 use Friendica\Util\Network;
  26 use Friendica\Util\XML;
  27 use DomXPath;
  28
  29 require_once 'include/dba.php';
  30
  31 /**
  32  * @brief This class contain functions for probing URL
  33  *
  34  */
  35 class Probe
  36 {
  37         private static $baseurl;
  38
  39         /**
  40          * @brief Rearrange the array so that it always has the same order
  41          *
  42          * @param array $data Unordered data
  43          *
  44          * @return array Ordered data
  45          */
  46         private static function rearrangeData($data)
  47         {
  48                 $fields = ["name", "nick", "guid", "url", "addr", "alias",
  49                                 "photo", "community", "keywords", "location", "about",
  50                                 "batch", "notify", "poll", "request", "confirm", "poco",
  51                                 "priority", "network", "pubkey", "baseurl"];
  52
  53                 $newdata = [];
  54                 foreach ($fields as $field) {
  55                         if (isset($data[$field])) {
  56                                 $newdata[$field] = $data[$field];
  57                         } else {
  58                                 $newdata[$field] = "";
  59                         }
  60                 }
  61
  62                 // We don't use the "priority" field anymore and replace it with a dummy.
  63                 $newdata["priority"] = 0;
  64
  65                 return $newdata;
  66         }
  67
  68         /**
  69          * @brief Check if the hostname belongs to the own server
  70          *
  71          * @param string $host The hostname that is to be checked
  72          *
  73          * @return bool Does the testes hostname belongs to the own server?
  74          */
  75         private static function ownHost($host)
  76         {
  77                 $own_host = get_app()->get_hostname();
  78
  79                 $parts = parse_url($host);
  80
  81                 if (!isset($parts['scheme'])) {
  82                         $parts = parse_url('http://'.$host);
  83                 }
  84
  85                 if (!isset($parts['host'])) {
  86                         return false;
  87                 }
  88                 return $parts['host'] == $own_host;
  89         }
  90
  91         /**
  92          * @brief Probes for webfinger path via "host-meta"
  93          *
  94          * We have to check if the servers in the future still will offer this.
  95          * It seems as if it was dropped from the standard.
  96          *
  97          * @param string $host The host part of an url
  98          *
  99          * @return array with template and type of the webfinger template for JSON or XML
 100          */
 101         private static function hostMeta($host)
 102         {
 103                 // Reset the static variable
 104                 self::$baseurl = '';
 105
 106                 $ssl_url = "https://".$host."/.well-known/host-meta";
 107                 $url = "http://".$host."/.well-known/host-meta";
 108
 109                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 110                 $redirects = 0;
 111
 112                 logger("Probing for ".$host, LOGGER_DEBUG);
 113                 $xrd = null;
 114
 115                 $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 116                 if ($ret['success']) {
 117                         $xml = $ret['body'];
 118                         $xrd = XML::parseString($xml, false);
 119                         $host_url = 'https://'.$host;
 120                 }
 121
 122                 if (!is_object($xrd)) {
 123                         $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 124                         if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 125                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 126                                 return false;
 127                         }
 128                         $xml = $ret['body'];
 129                         $xrd = XML::parseString($xml, false);
 130                         $host_url = 'http://'.$host;
 131                 }
 132                 if (!is_object($xrd)) {
 133                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 134                         return [];
 135                 }
 136
 137                 $links = XML::elementToArray($xrd);
 138                 if (!isset($links["xrd"]["link"])) {
 139                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 140                         return [];
 141                 }
 142
 143                 $lrdd = [];
 144                 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
 145                 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
 146                 // compared to the XML webfinger. So this is commented out by now.
 147                 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
 148
 149                 foreach ($links["xrd"]["link"] as $value => $link) {
 150                         if (!empty($link["@attributes"])) {
 151                                 $attributes = $link["@attributes"];
 152                         } elseif ($value == "@attributes") {
 153                                 $attributes = $link;
 154                         } else {
 155                                 continue;
 156                         }
 157
 158                         if (($attributes["rel"] == "lrdd") && !empty($attributes["template"])) {
 159                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 160
 161                                 $lrdd[$type] = $attributes["template"];
 162                         }
 163                 }
 164
 165                 self::$baseurl = "http://".$host;
 166
 167                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 168
 169                 return $lrdd;
 170         }
 171
 172         /**
 173          * @brief Perform Webfinger lookup and return DFRN data
 174          *
 175          * Given an email style address, perform webfinger lookup and
 176          * return the resulting DFRN profile URL, or if no DFRN profile URL
 177          * is located, returns an OStatus subscription template (prefixed
 178          * with the string 'stat:' to identify it as on OStatus template).
 179          * If this isn't an email style address just return $webbie.
 180          * Return an empty string if email-style addresses but webfinger fails,
 181          * or if the resultant personal XRD doesn't contain a supported
 182          * subscription/friend-request attribute.
 183          *
 184          * amended 7/9/2011 to return an hcard which could save potentially loading
 185          * a lengthy content page to scrape dfrn attributes
 186          *
 187          * @param string $webbie    Address that should be probed
 188          * @param string $hcard_url Link to the hcard - is returned by reference
 189          *
 190          * @return string profile link
 191          */
 192         public static function webfingerDfrn($webbie, &$hcard_url)
 193         {
 194                 $profile_link = '';
 195
 196                 $links = self::lrdd($webbie);
 197                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 198                 if (count($links)) {
 199                         foreach ($links as $link) {
 200                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 201                                         $profile_link = $link['@attributes']['href'];
 202                                 }
 203                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 204                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 205                                 }
 206                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 207                                         $hcard_url = $link['@attributes']['href'];
 208                                 }
 209                         }
 210                 }
 211                 return $profile_link;
 212         }
 213
 214         /**
 215          * @brief Check an URI for LRDD data
 216          *
 217          * this is a replacement for the "lrdd" function.
 218          * It isn't used in this class and has some redundancies in the code.
 219          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 220          *
 221          * @param string $uri Address that should be probed
 222          *
 223          * @return array uri data
 224          */
 225         public static function lrdd($uri)
 226         {
 227                 $lrdd = self::hostMeta($uri);
 228                 $webfinger = null;
 229
 230                 if (is_bool($lrdd)) {
 231                         return [];
 232                 }
 233
 234                 if (!$lrdd) {
 235                         $parts = @parse_url($uri);
 236                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
 237                                 return [];
 238                         }
 239
 240                         $host = $parts["host"];
 241                         if (!empty($parts["port"])) {
 242                                 $host .= ':'.$parts["port"];
 243                         }
 244
 245                         $path_parts = explode("/", trim($parts["path"], "/"));
 246
 247                         $nick = array_pop($path_parts);
 248
 249                         do {
 250                                 $lrdd = self::hostMeta($host);
 251                                 $host .= "/".array_shift($path_parts);
 252                         } while (!$lrdd && (sizeof($path_parts) > 0));
 253                 }
 254
 255                 if (!$lrdd) {
 256                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 257                         return [];
 258                 }
 259
 260                 foreach ($lrdd as $type => $template) {
 261                         if ($webfinger) {
 262                                 continue;
 263                         }
 264
 265                         $path = str_replace('{uri}', urlencode($uri), $template);
 266                         $webfinger = self::webfinger($path, $type);
 267
 268                         if (!$webfinger && (strstr($uri, "@"))) {
 269                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
 270                                 $webfinger = self::webfinger($path, $type);
 271                         }
 272
 273                         // Special treatment for Mastodon
 274                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 275                         // But the webfinger for this format fails.
 276                         if (!$webfinger && !empty($nick)) {
 277                                 // Mastodon uses a "@" as prefix for usernames in their url format
 278                                 $nick = ltrim($nick, '@');
 279
 280                                 $addr = $nick."@".$host;
 281
 282                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 283                                 $webfinger = self::webfinger($path, $type);
 284                         }
 285                 }
 286
 287                 if (!is_array($webfinger["links"])) {
 288                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 289                         return false;
 290                 }
 291
 292                 $data = [];
 293
 294                 foreach ($webfinger["links"] as $link) {
 295                         $data[] = ["@attributes" => $link];
 296                 }
 297
 298                 if (is_array($webfinger["aliases"])) {
 299                         foreach ($webfinger["aliases"] as $alias) {
 300                                 $data[] = ["@attributes" =>
 301                                                         ["rel" => "alias",
 302                                                                 "href" => $alias]];
 303                         }
 304                 }
 305
 306                 return $data;
 307         }
 308
 309         /**
 310          * @brief Fetch information (protocol endpoints and user information) about a given uri
 311          *
 312          * @param string  $uri     Address that should be probed
 313          * @param string  $network Test for this specific network
 314          * @param integer $uid     User ID for the probe (only used for mails)
 315          * @param boolean $cache   Use cached values?
 316          *
 317          * @return array uri data
 318          */
 319         public static function uri($uri, $network = "", $uid = -1, $cache = true)
 320         {
 321                 if ($cache) {
 322                         $result = Cache::get("Probe::uri:".$network.":".$uri);
 323                         if (!is_null($result)) {
 324                                 return $result;
 325                         }
 326                 }
 327
 328                 if ($uid == -1) {
 329                         $uid = local_user();
 330                 }
 331
 332                 if ($network != Protocol::ACTIVITYPUB) {
 333                         $data = self::detect($uri, $network, $uid);
 334                 } else {
 335                         $data = null;
 336                 }
 337
 338                 if (in_array(defaults($data, 'network', ''), ['', Protocol::PHANTOM])) {
 339                         $ap_profile = ActivityPub::probeProfile($uri);
 340                         if (!empty($ap_profile) && ($ap_profile['network'] == Protocol::ACTIVITYPUB)) {
 341                                 $data = $ap_profile;
 342                         }
 343                 }
 344
 345                 if (!isset($data["url"])) {
 346                         $data["url"] = $uri;
 347                 }
 348
 349                 if (x($data, "photo")) {
 350                         $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
 351                 } else {
 352                         $data["photo"] = System::baseUrl().'/images/person-175.jpg';
 353                 }
 354
 355                 if (empty($data["name"])) {
 356                         if (!empty($data["nick"])) {
 357                                 $data["name"] = $data["nick"];
 358                         }
 359
 360                         if (!x($data, "name")) {
 361                                 $data["name"] = $data["url"];
 362                         }
 363                 }
 364
 365                 if (empty($data["nick"])) {
 366                         $data["nick"] = strtolower($data["name"]);
 367
 368                         if (strpos($data['nick'], ' ')) {
 369                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 370                         }
 371                 }
 372
 373                 if (!empty(self::$baseurl)) {
 374                         $data["baseurl"] = self::$baseurl;
 375                 }
 376
 377                 if (empty($data["network"])) {
 378                         $data["network"] = Protocol::PHANTOM;
 379                 }
 380
 381                 $data = self::rearrangeData($data);
 382
 383                 // Only store into the cache if the value seems to be valid
 384                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
 385                         Cache::set("Probe::uri:".$network.":".$uri, $data, CACHE_DAY);
 386
 387                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 388                         /// The biggest problem is the avatar picture that could have a reduced image size.
 389                         /// It should only be updated if the existing picture isn't existing anymore.
 390                         /// We only update the contact when it is no probing for a specific network.
 391                         if (($data['network'] != Protocol::FEED)
 392                                 && ($network == "")
 393                                 && $data["name"]
 394                                 && $data["nick"]
 395                                 && $data["url"]
 396                                 && $data["addr"]
 397                                 && $data["poll"]
 398                         ) {
 399                                 $fields = ['name' => $data['name'],
 400                                                 'nick' => $data['nick'],
 401                                                 'url' => $data['url'],
 402                                                 'addr' => $data['addr'],
 403                                                 'photo' => $data['photo'],
 404                                                 'keywords' => $data['keywords'],
 405                                                 'location' => $data['location'],
 406                                                 'about' => $data['about'],
 407                                                 'notify' => $data['notify'],
 408                                                 'network' => $data['network'],
 409                                                 'server_url' => $data['baseurl']];
 410
 411                                 // This doesn't cover the case when a community isn't a community anymore
 412                                 if (!empty($data['community']) && $data['community']) {
 413                                         $fields['community'] = $data['community'];
 414                                         $fields['contact-type'] = Contact::ACCOUNT_TYPE_COMMUNITY;
 415                                 }
 416
 417                                 $fieldnames = [];
 418
 419                                 foreach ($fields as $key => $val) {
 420                                         if (empty($val)) {
 421                                                 unset($fields[$key]);
 422                                         } else {
 423                                                 $fieldnames[] = $key;
 424                                         }
 425                                 }
 426
 427                                 $fields['updated'] = DateTimeFormat::utcNow();
 428
 429                                 $condition = ['nurl' => normalise_link($data["url"])];
 430
 431                                 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
 432
 433                                 // When the gcontact doesn't exist, the value "true" will trigger an insert.
 434                                 // In difference to the public contacts we want to have every contact
 435                                 // in the world in our global contacts.
 436                                 if (!$old_fields) {
 437                                         $old_fields = true;
 438
 439                                         // These values have to be set only on insert
 440                                         $fields['photo'] = $data['photo'];
 441                                         $fields['created'] = DateTimeFormat::utcNow();
 442                                 }
 443
 444                                 DBA::update('gcontact', $fields, $condition, $old_fields);
 445
 446                                 $fields = ['name' => $data['name'],
 447                                                 'nick' => $data['nick'],
 448                                                 'url' => $data['url'],
 449                                                 'addr' => $data['addr'],
 450                                                 'alias' => $data['alias'],
 451                                                 'keywords' => $data['keywords'],
 452                                                 'location' => $data['location'],
 453                                                 'about' => $data['about'],
 454                                                 'batch' => $data['batch'],
 455                                                 'notify' => $data['notify'],
 456                                                 'poll' => $data['poll'],
 457                                                 'request' => $data['request'],
 458                                                 'confirm' => $data['confirm'],
 459                                                 'poco' => $data['poco'],
 460                                                 'network' => $data['network'],
 461                                                 'pubkey' => $data['pubkey'],
 462                                                 'priority' => $data['priority'],
 463                                                 'writable' => true,
 464                                                 'rel' => Contact::SHARING];
 465
 466                                 $fieldnames = [];
 467
 468                                 foreach ($fields as $key => $val) {
 469                                         if (empty($val)) {
 470                                                 unset($fields[$key]);
 471                                         } else {
 472                                                 $fieldnames[] = $key;
 473                                         }
 474                                 }
 475
 476                                 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 477
 478                                 // "$old_fields" will return a "false" when the contact doesn't exist.
 479                                 // This won't trigger an insert. This is intended, since we only need
 480                                 // public contacts for everyone we store items from.
 481                                 // We don't need to store every contact on the planet.
 482                                 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 483
 484                                 $fields['name-date'] = DateTimeFormat::utcNow();
 485                                 $fields['uri-date'] = DateTimeFormat::utcNow();
 486                                 $fields['success_update'] = DateTimeFormat::utcNow();
 487
 488                                 DBA::update('contact', $fields, $condition, $old_fields);
 489                         }
 490                 }
 491
 492                 return $data;
 493         }
 494
 495         /**
 496          * @brief Switch the scheme of an url between http and https
 497          *
 498          * @param string $url URL
 499          *
 500          * @return string switched URL
 501          */
 502         private static function switchScheme($url)
 503         {
 504                 $parts = parse_url($url);
 505
 506                 if (!isset($parts['scheme'])) {
 507                         return $url;
 508                 }
 509
 510                 if ($parts['scheme'] == 'http') {
 511                         $url = str_replace('http://', 'https://', $url);
 512                 } elseif ($parts['scheme'] == 'https') {
 513                         $url = str_replace('https://', 'http://', $url);
 514                 }
 515
 516                 return $url;
 517         }
 518
 519         /**
 520          * @brief Checks if a profile url should be OStatus but only provides partial information
 521          *
 522          * @param array  $webfinger Webfinger data
 523          * @param string $lrdd      Path template for webfinger request
 524          * @param string $type      type
 525          *
 526          * @return array fixed webfinger data
 527          */
 528         private static function fixOStatus($webfinger, $lrdd, $type)
 529         {
 530                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 531                         return $webfinger;
 532                 }
 533
 534                 $is_ostatus = false;
 535                 $has_key = false;
 536
 537                 foreach ($webfinger['links'] as $link) {
 538                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 539                                 $is_ostatus = true;
 540                         }
 541                         if ($link['rel'] == 'magic-public-key') {
 542                                 $has_key = true;
 543                         }
 544                 }
 545
 546                 if (!$is_ostatus || $has_key) {
 547                         return $webfinger;
 548                 }
 549
 550                 $url = self::switchScheme($webfinger['subject']);
 551                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 552                 $webfinger2 = self::webfinger($path, $type);
 553
 554                 // Is the new webfinger detectable as OStatus?
 555                 if (self::ostatus($webfinger2, true)) {
 556                         $webfinger = $webfinger2;
 557                 }
 558
 559                 return $webfinger;
 560         }
 561
 562         /**
 563          * @brief Fetch information (protocol endpoints and user information) about a given uri
 564          *
 565          * This function is only called by the "uri" function that adds caching and rearranging of data.
 566          *
 567          * @param string  $uri     Address that should be probed
 568          * @param string  $network Test for this specific network
 569          * @param integer $uid     User ID for the probe (only used for mails)
 570          *
 571          * @return array uri data
 572          */
 573         private static function detect($uri, $network, $uid)
 574         {
 575                 $parts = parse_url($uri);
 576
 577                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 578                         $host = $parts["host"];
 579                         if (!empty($parts["port"])) {
 580                                 $host .= ':'.$parts["port"];
 581                         }
 582
 583                         if ($host == 'twitter.com') {
 584                                 return ["network" => Protocol::TWITTER];
 585                         }
 586                         $lrdd = self::hostMeta($host);
 587
 588                         if (is_bool($lrdd)) {
 589                                 return [];
 590                         }
 591
 592                         $path_parts = explode("/", trim($parts["path"], "/"));
 593
 594                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 595                                 $host .= "/".array_shift($path_parts);
 596                                 $lrdd = self::hostMeta($host);
 597                         }
 598                         if (!$lrdd) {
 599                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 600                                 return self::feed($uri);
 601                         }
 602                         $nick = array_pop($path_parts);
 603
 604                         // Mastodon uses a "@" as prefix for usernames in their url format
 605                         $nick = ltrim($nick, '@');
 606
 607                         $addr = $nick."@".$host;
 608                 } elseif (strstr($uri, '@')) {
 609                         // If the URI starts with "mailto:" then jump directly to the mail detection
 610                         if (strpos($uri, 'mailto:') !== false) {
 611                                 $uri = str_replace('mailto:', '', $uri);
 612                                 return self::mail($uri, $uid);
 613                         }
 614
 615                         if ($network == Protocol::MAIL) {
 616                                 return self::mail($uri, $uid);
 617                         }
 618                         // Remove "acct:" from the URI
 619                         $uri = str_replace('acct:', '', $uri);
 620
 621                         $host = substr($uri, strpos($uri, '@') + 1);
 622                         $nick = substr($uri, 0, strpos($uri, '@'));
 623
 624                         if (strpos($uri, '@twitter.com')) {
 625                                 return ["network" => Protocol::TWITTER];
 626                         }
 627                         $lrdd = self::hostMeta($host);
 628
 629                         if (is_bool($lrdd)) {
 630                                 return [];
 631                         }
 632
 633                         if (!$lrdd) {
 634                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 635                                 return self::mail($uri, $uid);
 636                         }
 637                         $addr = $uri;
 638                 } else {
 639                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 640                         return false;
 641                 }
 642
 643                 $webfinger = false;
 644
 645                 /// @todo Do we need the prefix "acct:" or "acct://"?
 646
 647                 foreach ($lrdd as $type => $template) {
 648                         if ($webfinger) {
 649                                 continue;
 650                         }
 651
 652                         // At first try it with the given uri
 653                         $path = str_replace('{uri}', urlencode($uri), $template);
 654                         $webfinger = self::webfinger($path, $type);
 655
 656                         // Fix possible problems with GNU Social probing to wrong scheme
 657                         $webfinger = self::fixOStatus($webfinger, $template, $type);
 658
 659                         // We cannot be sure that the detected address was correct, so we don't use the values
 660                         if ($webfinger && ($uri != $addr)) {
 661                                 $nick = "";
 662                                 $addr = "";
 663                         }
 664
 665                         // Try webfinger with the address (user@domain.tld)
 666                         if (!$webfinger) {
 667                                 $path = str_replace('{uri}', urlencode($addr), $template);
 668                                 $webfinger = self::webfinger($path, $type);
 669                         }
 670
 671                         // Mastodon needs to have it with "acct:"
 672                         if (!$webfinger) {
 673                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
 674                                 $webfinger = self::webfinger($path, $type);
 675                         }
 676                 }
 677
 678                 if (!$webfinger) {
 679                         return self::feed($uri);
 680                 }
 681
 682                 $result = false;
 683
 684                 logger("Probing ".$uri, LOGGER_DEBUG);
 685
 686                 if (in_array($network, ["", Protocol::DFRN])) {
 687                         $result = self::dfrn($webfinger);
 688                 }
 689                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 690                         $result = self::diaspora($webfinger);
 691                 }
 692                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 693                         $result = self::ostatus($webfinger);
 694                 }
 695                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 696                         $result = self::pumpio($webfinger, $addr);
 697                 }
 698                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
 699                         $result = self::feed($uri);
 700                 } else {
 701                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 702                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 703                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 704                                 $result["nick"] = $nick;
 705                         }
 706
 707                         if (empty($result["addr"]) && ($addr != "")) {
 708                                 $result["addr"] = $addr;
 709                         }
 710                 }
 711
 712                 if (empty($result["network"])) {
 713                         $result["network"] = Protocol::PHANTOM;
 714                 }
 715
 716                 if (empty($result["url"])) {
 717                         $result["url"] = $uri;
 718                 }
 719
 720                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 721
 722                 if (empty($result["baseurl"])) {
 723                         $pos = strpos($result["url"], $host);
 724                         if ($pos) {
 725                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 726                         }
 727                 }
 728                 return $result;
 729         }
 730
 731         /**
 732          * @brief Perform a webfinger request.
 733          *
 734          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 735          *
 736          * @param string $url  Address that should be probed
 737          * @param string $type type
 738          *
 739          * @return array webfinger data
 740          */
 741         private static function webfinger($url, $type)
 742         {
 743                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 744                 $redirects = 0;
 745
 746                 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 747                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 748                         return false;
 749                 }
 750                 $data = $ret['body'];
 751
 752                 $webfinger = json_decode($data, true);
 753                 if (is_array($webfinger)) {
 754                         if (!isset($webfinger["links"])) {
 755                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 756                                 return false;
 757                         }
 758                         return $webfinger;
 759                 }
 760
 761                 // If it is not JSON, maybe it is XML
 762                 $xrd = XML::parseString($data, false);
 763                 if (!is_object($xrd)) {
 764                         logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
 765                         return false;
 766                 }
 767
 768                 $xrd_arr = XML::elementToArray($xrd);
 769                 if (!isset($xrd_arr["xrd"]["link"])) {
 770                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 771                         return false;
 772                 }
 773
 774                 $webfinger = [];
 775
 776                 if (!empty($xrd_arr["xrd"]["subject"])) {
 777                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 778                 }
 779
 780                 if (!empty($xrd_arr["xrd"]["alias"])) {
 781                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 782                 }
 783
 784                 $webfinger["links"] = [];
 785
 786                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 787                         if (!empty($data["@attributes"])) {
 788                                 $attributes = $data["@attributes"];
 789                         } elseif ($value == "@attributes") {
 790                                 $attributes = $data;
 791                         } else {
 792                                 continue;
 793                         }
 794
 795                         $webfinger["links"][] = $attributes;
 796                 }
 797                 return $webfinger;
 798         }
 799
 800         /**
 801          * @brief Poll the Friendica specific noscrape page.
 802          *
 803          * "noscrape" is a faster alternative to fetch the data from the hcard.
 804          * This functionality was originally created for the directory.
 805          *
 806          * @param string $noscrape_url Link to the noscrape page
 807          * @param array  $data         The already fetched data
 808          *
 809          * @return array noscrape data
 810          */
 811         private static function pollNoscrape($noscrape_url, $data)
 812         {
 813                 $ret = Network::curl($noscrape_url);
 814                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
 815                         return false;
 816                 }
 817                 $content = $ret['body'];
 818                 if (!$content) {
 819                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 820                         return false;
 821                 }
 822
 823                 $json = json_decode($content, true);
 824                 if (!is_array($json)) {
 825                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 826                         return false;
 827                 }
 828
 829                 if (!empty($json["fn"])) {
 830                         $data["name"] = $json["fn"];
 831                 }
 832
 833                 if (!empty($json["addr"])) {
 834                         $data["addr"] = $json["addr"];
 835                 }
 836
 837                 if (!empty($json["nick"])) {
 838                         $data["nick"] = $json["nick"];
 839                 }
 840
 841                 if (!empty($json["guid"])) {
 842                         $data["guid"] = $json["guid"];
 843                 }
 844
 845                 if (!empty($json["comm"])) {
 846                         $data["community"] = $json["comm"];
 847                 }
 848
 849                 if (!empty($json["tags"])) {
 850                         $keywords = implode(" ", $json["tags"]);
 851                         if ($keywords != "") {
 852                                 $data["keywords"] = $keywords;
 853                         }
 854                 }
 855
 856                 $location = Profile::formatLocation($json);
 857                 if ($location) {
 858                         $data["location"] = $location;
 859                 }
 860
 861                 if (!empty($json["about"])) {
 862                         $data["about"] = $json["about"];
 863                 }
 864
 865                 if (!empty($json["key"])) {
 866                         $data["pubkey"] = $json["key"];
 867                 }
 868
 869                 if (!empty($json["photo"])) {
 870                         $data["photo"] = $json["photo"];
 871                 }
 872
 873                 if (!empty($json["dfrn-request"])) {
 874                         $data["request"] = $json["dfrn-request"];
 875                 }
 876
 877                 if (!empty($json["dfrn-confirm"])) {
 878                         $data["confirm"] = $json["dfrn-confirm"];
 879                 }
 880
 881                 if (!empty($json["dfrn-notify"])) {
 882                         $data["notify"] = $json["dfrn-notify"];
 883                 }
 884
 885                 if (!empty($json["dfrn-poll"])) {
 886                         $data["poll"] = $json["dfrn-poll"];
 887                 }
 888
 889                 return $data;
 890         }
 891
 892         /**
 893          * @brief Check for valid DFRN data
 894          *
 895          * @param array $data DFRN data
 896          *
 897          * @return int Number of errors
 898          */
 899         public static function validDfrn($data)
 900         {
 901                 $errors = 0;
 902                 if (!isset($data['key'])) {
 903                         $errors ++;
 904                 }
 905                 if (!isset($data['dfrn-request'])) {
 906                         $errors ++;
 907                 }
 908                 if (!isset($data['dfrn-confirm'])) {
 909                         $errors ++;
 910                 }
 911                 if (!isset($data['dfrn-notify'])) {
 912                         $errors ++;
 913                 }
 914                 if (!isset($data['dfrn-poll'])) {
 915                         $errors ++;
 916                 }
 917                 return $errors;
 918         }
 919
 920         /**
 921          * @brief Fetch data from a DFRN profile page and via "noscrape"
 922          *
 923          * @param string $profile_link Link to the profile page
 924          *
 925          * @return array profile data
 926          */
 927         public static function profile($profile_link)
 928         {
 929                 $data = [];
 930
 931                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 932
 933                 // Fetch data via noscrape - this is faster
 934                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
 935                 $data = self::pollNoscrape($noscrape_url, $data);
 936
 937                 if (!isset($data["notify"])
 938                         || !isset($data["confirm"])
 939                         || !isset($data["request"])
 940                         || !isset($data["poll"])
 941                         || !isset($data["name"])
 942                         || !isset($data["photo"])
 943                 ) {
 944                         $data = self::pollHcard($profile_link, $data, true);
 945                 }
 946
 947                 $prof_data = [];
 948
 949                 if (empty($data["addr"]) || empty($data["nick"])) {
 950                         $probe_data = self::uri($profile_link);
 951                         $data["addr"] = defaults($data, "addr", $probe_data["addr"]);
 952                         $data["nick"] = defaults($data, "nick", $probe_data["nick"]);
 953                 }
 954
 955                 $prof_data["addr"]         = $data["addr"];
 956                 $prof_data["nick"]         = $data["nick"];
 957                 $prof_data["dfrn-request"] = $data["request"];
 958                 $prof_data["dfrn-confirm"] = $data["confirm"];
 959                 $prof_data["dfrn-notify"]  = $data["notify"];
 960                 $prof_data["dfrn-poll"]    = $data["poll"];
 961                 $prof_data["photo"]        = $data["photo"];
 962                 $prof_data["fn"]           = $data["name"];
 963                 $prof_data["key"]          = $data["pubkey"];
 964
 965                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 966
 967                 return $prof_data;
 968         }
 969
 970         /**
 971          * @brief Check for DFRN contact
 972          *
 973          * @param array $webfinger Webfinger data
 974          *
 975          * @return array DFRN data
 976          */
 977         private static function dfrn($webfinger)
 978         {
 979                 $hcard_url = "";
 980                 $data = [];
 981                 foreach ($webfinger["links"] as $link) {
 982                         if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) {
 983                                 $data["network"] = Protocol::DFRN;
 984                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
 985                                 $data["poll"] = $link["href"];
 986                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
 987                                 $data["url"] = $link["href"];
 988                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
 989                                 $hcard_url = $link["href"];
 990                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
 991                                 $data["poco"] = $link["href"];
 992                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
 993                                 $data["photo"] = $link["href"];
 994                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
 995                                 $data["baseurl"] = trim($link["href"], '/');
 996                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
 997                                 $data["guid"] = $link["href"];
 998                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
 999                                 $data["pubkey"] = base64_decode($link["href"]);
1000
1001                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1002                                 if (strstr($data["pubkey"], 'RSA ')) {
1003                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1004                                 }
1005                         }
1006                 }
1007
1008                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1009                         foreach ($webfinger["aliases"] as $alias) {
1010                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1011                                         $data["url"] = $alias;
1012                                 } elseif (!strstr($alias, "@") && normalise_link($alias) != normalise_link($data["url"])) {
1013                                         $data["alias"] = $alias;
1014                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1015                                         $data["addr"] = substr($alias, 5);
1016                                 }
1017                         }
1018                 }
1019
1020                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1021                         $data["addr"] = substr($webfinger["subject"], 5);
1022                 }
1023
1024                 if (!isset($data["network"]) || ($hcard_url == "")) {
1025                         return false;
1026                 }
1027
1028                 // Fetch data via noscrape - this is faster
1029                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1030                 $data = self::pollNoscrape($noscrape_url, $data);
1031
1032                 if (isset($data["notify"])
1033                         && isset($data["confirm"])
1034                         && isset($data["request"])
1035                         && isset($data["poll"])
1036                         && isset($data["name"])
1037                         && isset($data["photo"])
1038                 ) {
1039                         return $data;
1040                 }
1041
1042                 $data = self::pollHcard($hcard_url, $data, true);
1043
1044                 return $data;
1045         }
1046
1047         /**
1048          * @brief Poll the hcard page (Diaspora and Friendica specific)
1049          *
1050          * @param string  $hcard_url Link to the hcard page
1051          * @param array   $data      The already fetched data
1052          * @param boolean $dfrn      Poll DFRN specific data
1053          *
1054          * @return array hcard data
1055          */
1056         private static function pollHcard($hcard_url, $data, $dfrn = false)
1057         {
1058                 $ret = Network::curl($hcard_url);
1059                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1060                         return false;
1061                 }
1062                 $content = $ret['body'];
1063                 if (!$content) {
1064                         return false;
1065                 }
1066
1067                 $doc = new DOMDocument();
1068                 if (!@$doc->loadHTML($content)) {
1069                         return false;
1070                 }
1071
1072                 $xpath = new DomXPath($doc);
1073
1074                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1075                 if (!is_object($vcards)) {
1076                         return false;
1077                 }
1078
1079                 if (!isset($data["baseurl"])) {
1080                         $data["baseurl"] = "";
1081                 }
1082
1083                 if ($vcards->length > 0) {
1084                         $vcard = $vcards->item(0);
1085
1086                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1087                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1088                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1089                         if (($search->length > 0) && empty($data["guid"])) {
1090                                 $data["guid"] = $search->item(0)->nodeValue;
1091                         }
1092
1093                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1094                         if ($search->length > 0) {
1095                                 $data["nick"] = $search->item(0)->nodeValue;
1096                         }
1097
1098                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1099                         if ($search->length > 0) {
1100                                 $data["name"] = $search->item(0)->nodeValue;
1101                         }
1102
1103                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1104                         if ($search->length > 0) {
1105                                 $data["searchable"] = $search->item(0)->nodeValue;
1106                         }
1107
1108                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1109                         if ($search->length > 0) {
1110                                 $data["pubkey"] = $search->item(0)->nodeValue;
1111                                 if (strstr($data["pubkey"], 'RSA ')) {
1112                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1113                                 }
1114                         }
1115
1116                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1117                         if ($search->length > 0) {
1118                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1119                         }
1120                 }
1121
1122                 $avatar = [];
1123                 if (!empty($vcard)) {
1124                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1125                         foreach ($photos as $photo) {
1126                                 $attr = [];
1127                                 foreach ($photo->attributes as $attribute) {
1128                                         $attr[$attribute->name] = trim($attribute->value);
1129                                 }
1130
1131                                 if (isset($attr["src"]) && isset($attr["width"])) {
1132                                         $avatar[$attr["width"]] = $attr["src"];
1133                                 }
1134
1135                                 // We don't have a width. So we just take everything that we got.
1136                                 // This is a Hubzilla workaround which doesn't send a width.
1137                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1138                                         $avatar[] = $attr["src"];
1139                                 }
1140                         }
1141                 }
1142
1143                 if (sizeof($avatar)) {
1144                         ksort($avatar);
1145                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1146                 }
1147
1148                 if ($dfrn) {
1149                         // Poll DFRN specific data
1150                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1151                         if ($search->length > 0) {
1152                                 foreach ($search as $link) {
1153                                         //$data["request"] = $search->item(0)->nodeValue;
1154                                         $attr = [];
1155                                         foreach ($link->attributes as $attribute) {
1156                                                 $attr[$attribute->name] = trim($attribute->value);
1157                                         }
1158
1159                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1160                                 }
1161                         }
1162
1163                         // Older Friendica versions had used the "uid" field differently than newer versions
1164                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1165                                 unset($data["guid"]);
1166                         }
1167                 }
1168
1169
1170                 return $data;
1171         }
1172
1173         /**
1174          * @brief Check for Diaspora contact
1175          *
1176          * @param array $webfinger Webfinger data
1177          *
1178          * @return array Diaspora data
1179          */
1180         private static function diaspora($webfinger)
1181         {
1182                 $hcard_url = "";
1183                 $data = [];
1184                 foreach ($webfinger["links"] as $link) {
1185                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1186                                 $hcard_url = $link["href"];
1187                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1188                                 $data["baseurl"] = trim($link["href"], '/');
1189                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1190                                 $data["guid"] = $link["href"];
1191                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (defaults($link, "type", "") == "text/html") && !empty($link["href"])) {
1192                                 $data["url"] = $link["href"];
1193                         } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1194                                 $data["poll"] = $link["href"];
1195                         } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) {
1196                                 $data["poco"] = $link["href"];
1197                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1198                                 $data["notify"] = $link["href"];
1199                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1200                                 $data["pubkey"] = base64_decode($link["href"]);
1201
1202                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1203                                 if (strstr($data["pubkey"], 'RSA ')) {
1204                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1205                                 }
1206                         }
1207                 }
1208
1209                 if (!isset($data["url"]) || ($hcard_url == "")) {
1210                         return false;
1211                 }
1212
1213                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1214                         foreach ($webfinger["aliases"] as $alias) {
1215                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1216                                         $data["alias"] = $alias;
1217                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1218                                         $data["addr"] = substr($alias, 5);
1219                                 }
1220                         }
1221                 }
1222
1223                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1224                         $data["addr"] = substr($webfinger["subject"], 5);
1225                 }
1226
1227                 // Fetch further information from the hcard
1228                 $data = self::pollHcard($hcard_url, $data);
1229
1230                 if (!$data) {
1231                         return false;
1232                 }
1233
1234                 if (isset($data["url"])
1235                         && isset($data["guid"])
1236                         && isset($data["baseurl"])
1237                         && isset($data["pubkey"])
1238                         && ($hcard_url != "")
1239                 ) {
1240                         $data["network"] = Protocol::DIASPORA;
1241
1242                         // The Diaspora handle must always be lowercase
1243                         if (!empty($data["addr"])) {
1244                                 $data["addr"] = strtolower($data["addr"]);
1245                         }
1246
1247                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1248                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1249                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1250                 } else {
1251                         return false;
1252                 }
1253
1254                 return $data;
1255         }
1256
1257         /**
1258          * @brief Check for OStatus contact
1259          *
1260          * @param array $webfinger Webfinger data
1261          * @param bool  $short     Short detection mode
1262          *
1263          * @return array|bool OStatus data or "false" on error or "true" on short mode
1264          */
1265         private static function ostatus($webfinger, $short = false)
1266         {
1267                 $data = [];
1268
1269                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1270                         foreach ($webfinger["aliases"] as $alias) {
1271                                 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1272                                         $data["addr"] = str_replace('acct:', '', $alias);
1273                                 }
1274                         }
1275                 }
1276
1277                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1278                         && !strstr(normalise_link($webfinger["subject"]), "http://")
1279                 ) {
1280                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1281                 }
1282
1283                 $pubkey = "";
1284                 if (is_array($webfinger["links"])) {
1285                         foreach ($webfinger["links"] as $link) {
1286                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1287                                         && (defaults($link, "type", "") == "text/html")
1288                                         && ($link["href"] != "")
1289                                 ) {
1290                                         $data["url"] = $link["href"];
1291                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1292                                         $data["notify"] = $link["href"];
1293                                 } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) {
1294                                         $data["poll"] = $link["href"];
1295                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1296                                         $pubkey = $link["href"];
1297
1298                                         if (substr($pubkey, 0, 5) === 'data:') {
1299                                                 if (strstr($pubkey, ',')) {
1300                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1301                                                 } else {
1302                                                         $pubkey = substr($pubkey, 5);
1303                                                 }
1304                                         } elseif (normalise_link($pubkey) == 'http://') {
1305                                                 $ret = Network::curl($pubkey);
1306                                                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1307                                                         return false;
1308                                                 }
1309                                                 $pubkey = $ret['body'];
1310                                         }
1311
1312                                         $key = explode(".", $pubkey);
1313
1314                                         if (sizeof($key) >= 3) {
1315                                                 $m = base64url_decode($key[1]);
1316                                                 $e = base64url_decode($key[2]);
1317                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1318                                         }
1319                                 }
1320                         }
1321                 }
1322
1323                 if (isset($data["notify"]) && isset($data["pubkey"])
1324                         && isset($data["poll"])
1325                         && isset($data["url"])
1326                 ) {
1327                         $data["network"] = Protocol::OSTATUS;
1328                 } else {
1329                         return false;
1330                 }
1331
1332                 if ($short) {
1333                         return true;
1334                 }
1335
1336                 // Fetch all additional data from the feed
1337                 $ret = Network::curl($data["poll"]);
1338                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1339                         return false;
1340                 }
1341                 $feed = $ret['body'];
1342                 $dummy1 = null;
1343                 $dummy2 = null;
1344                 $dummy2 = null;
1345                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1346                 if (!$feed_data) {
1347                         return false;
1348                 }
1349
1350                 if (!empty($feed_data["header"]["author-name"])) {
1351                         $data["name"] = $feed_data["header"]["author-name"];
1352                 }
1353                 if (!empty($feed_data["header"]["author-nick"])) {
1354                         $data["nick"] = $feed_data["header"]["author-nick"];
1355                 }
1356                 if (!empty($feed_data["header"]["author-avatar"])) {
1357                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1358                 }
1359                 if (!empty($feed_data["header"]["author-id"])) {
1360                         $data["alias"] = $feed_data["header"]["author-id"];
1361                 }
1362                 if (!empty($feed_data["header"]["author-location"])) {
1363                         $data["location"] = $feed_data["header"]["author-location"];
1364                 }
1365                 if (!empty($feed_data["header"]["author-about"])) {
1366                         $data["about"] = $feed_data["header"]["author-about"];
1367                 }
1368                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1369                 // So we take the value that we just fetched, although the other one worked as well
1370                 if (!empty($feed_data["header"]["author-link"])) {
1371                         $data["url"] = $feed_data["header"]["author-link"];
1372                 }
1373
1374                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1375                         $data['url'] = $data["alias"];
1376                         $data["alias"] = '';
1377                 }
1378
1379                 /// @todo Fetch location and "about" from the feed as well
1380                 return $data;
1381         }
1382
1383         /**
1384          * @brief Fetch data from a pump.io profile page
1385          *
1386          * @param string $profile_link Link to the profile page
1387          *
1388          * @return array profile data
1389          */
1390         private static function pumpioProfileData($profile_link)
1391         {
1392                 $doc = new DOMDocument();
1393                 if (!@$doc->loadHTMLFile($profile_link)) {
1394                         return false;
1395                 }
1396
1397                 $xpath = new DomXPath($doc);
1398
1399                 $data = [];
1400
1401                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1402
1403                 if ($data["name"] == '') {
1404                         // This is ugly - but pump.io doesn't seem to know a better way for it
1405                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1406                         $pos = strpos($data["name"], chr(10));
1407                         if ($pos) {
1408                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1409                         }
1410                 }
1411
1412                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1413
1414                 if ($data["location"] == '') {
1415                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1416                 }
1417
1418                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1419
1420                 if ($data["about"] == '') {
1421                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1422                 }
1423
1424                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1425                 if (!$avatar) {
1426                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1427                 }
1428                 if ($avatar) {
1429                         foreach ($avatar->attributes as $attribute) {
1430                                 if ($attribute->name == "src") {
1431                                         $data["photo"] = trim($attribute->value);
1432                                 }
1433                         }
1434                 }
1435
1436                 return $data;
1437         }
1438
1439         /**
1440          * @brief Check for pump.io contact
1441          *
1442          * @param array $webfinger Webfinger data
1443          *
1444          * @return array pump.io data
1445          */
1446         private static function pumpio($webfinger, $addr)
1447         {
1448                 $data = [];
1449                 foreach ($webfinger["links"] as $link) {
1450                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1451                                 && (defaults($link, "type", "") == "text/html")
1452                                 && ($link["href"] != "")
1453                         ) {
1454                                 $data["url"] = $link["href"];
1455                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1456                                 $data["notify"] = $link["href"];
1457                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1458                                 $data["poll"] = $link["href"];
1459                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1460                                 $data["dialback"] = $link["href"];
1461                         }
1462                 }
1463                 if (isset($data["poll"]) && isset($data["notify"])
1464                         && isset($data["dialback"])
1465                         && isset($data["url"])
1466                 ) {
1467                         // by now we use these fields only for the network type detection
1468                         // So we unset all data that isn't used at the moment
1469                         unset($data["dialback"]);
1470
1471                         $data["network"] = Protocol::PUMPIO;
1472                 } else {
1473                         return false;
1474                 }
1475
1476                 $profile_data = self::pumpioProfileData($data["url"]);
1477
1478                 if (!$profile_data) {
1479                         return false;
1480                 }
1481
1482                 $data = array_merge($data, $profile_data);
1483
1484                 if (($addr != '') && ($data['name'] != '')) {
1485                         $name = trim(str_replace($addr, '', $data['name']));
1486                         if ($name != '') {
1487                                 $data['name'] = $name;
1488                         }
1489                 }
1490
1491                 return $data;
1492         }
1493
1494         /**
1495          * @brief Check page for feed link
1496          *
1497          * @param string $url Page link
1498          *
1499          * @return string feed link
1500          */
1501         private static function getFeedLink($url)
1502         {
1503                 $doc = new DOMDocument();
1504
1505                 if (!@$doc->loadHTMLFile($url)) {
1506                         return false;
1507                 }
1508
1509                 $xpath = new DomXPath($doc);
1510
1511                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1512                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1513                 if (!is_object($feeds)) {
1514                         return false;
1515                 }
1516
1517                 if ($feeds->length == 0) {
1518                         return false;
1519                 }
1520
1521                 $feed_url = "";
1522
1523                 foreach ($feeds as $feed) {
1524                         $attr = [];
1525                         foreach ($feed->attributes as $attribute) {
1526                                 $attr[$attribute->name] = trim($attribute->value);
1527                         }
1528
1529                         if ($feed_url == "") {
1530                                 $feed_url = $attr["href"];
1531                         }
1532                 }
1533
1534                 return $feed_url;
1535         }
1536
1537         /**
1538          * @brief Check for feed contact
1539          *
1540          * @param string  $url   Profile link
1541          * @param boolean $probe Do a probe if the page contains a feed link
1542          *
1543          * @return array feed data
1544          */
1545         private static function feed($url, $probe = true)
1546         {
1547                 $ret = Network::curl($url);
1548                 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1549                         return false;
1550                 }
1551                 $feed = $ret['body'];
1552                 $dummy1 = $dummy2 = $dummy3 = null;
1553                 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1554
1555                 if (!$feed_data) {
1556                         if (!$probe) {
1557                                 return false;
1558                         }
1559
1560                         $feed_url = self::getFeedLink($url);
1561
1562                         if (!$feed_url) {
1563                                 return false;
1564                         }
1565
1566                         return self::feed($feed_url, false);
1567                 }
1568
1569                 if (!empty($feed_data["header"]["author-name"])) {
1570                         $data["name"] = $feed_data["header"]["author-name"];
1571                 }
1572
1573                 if (!empty($feed_data["header"]["author-nick"])) {
1574                         $data["nick"] = $feed_data["header"]["author-nick"];
1575                 }
1576
1577                 if (!empty($feed_data["header"]["author-avatar"])) {
1578                         $data["photo"] = $feed_data["header"]["author-avatar"];
1579                 }
1580
1581                 if (!empty($feed_data["header"]["author-id"])) {
1582                         $data["alias"] = $feed_data["header"]["author-id"];
1583                 }
1584
1585                 $data["url"] = $url;
1586                 $data["poll"] = $url;
1587
1588                 if (!empty($feed_data["header"]["author-link"])) {
1589                         $data["baseurl"] = $feed_data["header"]["author-link"];
1590                 } else {
1591                         $data["baseurl"] = $data["url"];
1592                 }
1593
1594                 $data["network"] = Protocol::FEED;
1595
1596                 return $data;
1597         }
1598
1599         /**
1600          * @brief Check for mail contact
1601          *
1602          * @param string  $uri Profile link
1603          * @param integer $uid User ID
1604          *
1605          * @return array mail data
1606          */
1607         private static function mail($uri, $uid)
1608         {
1609                 if (!Network::isEmailDomainValid($uri)) {
1610                         return false;
1611                 }
1612
1613                 if ($uid == 0) {
1614                         return false;
1615                 }
1616
1617                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1618
1619                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1620                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1621                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1622
1623                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1624                         return false;
1625                 }
1626
1627                 $mailbox = Email::constructMailboxName($mailacct);
1628                 $password = '';
1629                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1630                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1631                 if (!$mbox) {
1632                         return false;
1633                 }
1634
1635                 $msgs = Email::poll($mbox, $uri);
1636                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1637
1638                 if (!count($msgs)) {
1639                         return false;
1640                 }
1641
1642                 $phost = substr($uri, strpos($uri, '@') + 1);
1643
1644                 $data = [];
1645                 $data["addr"]    = $uri;
1646                 $data["network"] = Protocol::MAIL;
1647                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1648                 $data["nick"]    = $data["name"];
1649                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1650                 $data["url"]     = 'mailto:'.$uri;
1651                 $data["notify"]  = 'smtp '.random_string();
1652                 $data["poll"]    = 'email '.random_string();
1653
1654                 $x = Email::messageMeta($mbox, $msgs[0]);
1655                 if (stristr($x[0]->from, $uri)) {
1656                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1657                 } elseif (stristr($x[0]->to, $uri)) {
1658                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1659                 }
1660                 if (isset($adr)) {
1661                         foreach ($adr as $feadr) {
1662                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1663                                         &&(strcasecmp($feadr->host, $phost) == 0)
1664                                         && (strlen($feadr->personal))
1665                                 ) {
1666                                         $personal = imap_mime_header_decode($feadr->personal);
1667                                         $data["name"] = "";
1668                                         foreach ($personal as $perspart) {
1669                                                 if ($perspart->charset != "default") {
1670                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1671                                                 } else {
1672                                                         $data["name"] .= $perspart->text;
1673                                                 }
1674                                         }
1675
1676                                         $data["name"] = notags($data["name"]);
1677                                 }
1678                         }
1679                 }
1680                 if (!empty($mbox)) {
1681                         imap_close($mbox);
1682                 }
1683                 return $data;
1684         }
1685
1686         /**
1687          * @brief Mix two paths together to possibly fix missing parts
1688          *
1689          * @param string $avatar Path to the avatar
1690          * @param string $base   Another path that is hopefully complete
1691          *
1692          * @return string fixed avatar path
1693          */
1694         public static function fixAvatar($avatar, $base)
1695         {
1696                 $base_parts = parse_url($base);
1697
1698                 // Remove all parts that could create a problem
1699                 unset($base_parts['path']);
1700                 unset($base_parts['query']);
1701                 unset($base_parts['fragment']);
1702
1703                 $avatar_parts = parse_url($avatar);
1704
1705                 // Now we mix them
1706                 $parts = array_merge($base_parts, $avatar_parts);
1707
1708                 // And put them together again
1709                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1710                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1711                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1712                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1713                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1714                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1715
1716                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1717
1718                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1719
1720                 return $fixed;
1721         }
1722 }