include/Probe.php

   1 <?php
   2 /**
   3  * @file include/Probe.php
   4  * @brief Functions for probing URL
   5  *
   6  */
   7
   8 use \Friendica\Core\Config;
   9 use \Friendica\Core\PConfig;
  10
  11 require_once("include/feed.php");
  12 require_once('include/email.php');
  13 require_once('include/network.php');
  14
  15 /**
  16  * @brief This class contain functions for probing URL
  17  *
  18  */
  19 class Probe {
  20
  21         /**
  22          * @brief Rearrange the array so that it always has the same order
  23          *
  24          * @param array $data Unordered data
  25          *
  26          * @return array Ordered data
  27          */
  28         private function rearrange_data($data) {
  29                 $fields = array("name", "nick", "guid", "url", "addr", "alias",
  30                                 "photo", "community", "keywords", "location", "about",
  31                                 "batch", "notify", "poll", "request", "confirm", "poco",
  32                                 "priority", "network", "pubkey", "baseurl");
  33
  34                 $newdata = array();
  35                 foreach ($fields AS $field)
  36                         if (isset($data[$field]))
  37                                 $newdata[$field] = $data[$field];
  38                         else
  39                                 $newdata[$field] = "";
  40
  41                 // We don't use the "priority" field anymore and replace it with a dummy.
  42                 $newdata["priority"] = 0;
  43
  44                 return $newdata;
  45         }
  46
  47         /**
  48          * @brief Probes for XRD data
  49          *
  50          * @return array
  51          *      'lrdd' => Link to LRDD endpoint
  52          *      'lrdd-xml' => Link to LRDD endpoint in XML format
  53          *      'lrdd-json' => Link to LRDD endpoint in JSON format
  54          */
  55         private function xrd($host) {
  56
  57                 $ssl_url = "https://".$host."/.well-known/host-meta";
  58                 $url = "http://".$host."/.well-known/host-meta";
  59
  60                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
  61                 $redirects = 0;
  62
  63                 $ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
  64                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
  65                         return false;
  66                 }
  67                 $xml = $ret['body'];
  68
  69                 $xrd = parse_xml_string($xml, false);
  70
  71                 if (!is_object($xrd)) {
  72                         $xml = fetch_url($url, false, $redirects, $xrd_timeout, "application/xrd+xml");
  73                         $xrd = parse_xml_string($xml, false);
  74                 }
  75                 if (!is_object($xrd))
  76                         return false;
  77
  78                 $links = xml::element_to_array($xrd);
  79                 if (!isset($links["xrd"]["link"]))
  80                         return false;
  81
  82                 $xrd_data = array();
  83
  84                 foreach ($links["xrd"]["link"] AS $value => $link) {
  85                         if (isset($link["@attributes"]))
  86                                 $attributes = $link["@attributes"];
  87                         elseif ($value == "@attributes")
  88                                 $attributes = $link;
  89                         else
  90                                 continue;
  91
  92                         if (($attributes["rel"] == "lrdd") AND
  93                                 ($attributes["type"] == "application/xrd+xml"))
  94                                 $xrd_data["lrdd-xml"] = $attributes["template"];
  95                         elseif (($attributes["rel"] == "lrdd") AND
  96                                 ($attributes["type"] == "application/json"))
  97                                 $xrd_data["lrdd-json"] = $attributes["template"];
  98                         elseif ($attributes["rel"] == "lrdd")
  99                                 $xrd_data["lrdd"] = $attributes["template"];
 100                 }
 101                 return $xrd_data;
 102         }
 103
 104         /**
 105          * @brief Perform Webfinger lookup and return DFRN data
 106          *
 107          * Given an email style address, perform webfinger lookup and
 108          * return the resulting DFRN profile URL, or if no DFRN profile URL
 109          * is located, returns an OStatus subscription template (prefixed
 110          * with the string 'stat:' to identify it as on OStatus template).
 111          * If this isn't an email style address just return $webbie.
 112          * Return an empty string if email-style addresses but webfinger fails,
 113          * or if the resultant personal XRD doesn't contain a supported
 114          * subscription/friend-request attribute.
 115          *
 116          * amended 7/9/2011 to return an hcard which could save potentially loading
 117          * a lengthy content page to scrape dfrn attributes
 118          *
 119          * @param string $webbie Address that should be probed
 120          * @param string $hcard Link to the hcard - is returned by reference
 121          *
 122          * @return string profile link
 123          */
 124
 125         public static function webfinger_dfrn($webbie, &$hcard) {
 126
 127                 $profile_link = '';
 128
 129                 $links = self::lrdd($webbie);
 130                 logger('webfinger_dfrn: '.$webbie.':'.print_r($links,true), LOGGER_DATA);
 131                 if (count($links)) {
 132                         foreach ($links as $link) {
 133                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN)
 134                                         $profile_link = $link['@attributes']['href'];
 135                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) AND ($profile_link == ""))
 136                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 137                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 138                                         $hcard = $link['@attributes']['href'];
 139                         }
 140                 }
 141                 return $profile_link;
 142         }
 143
 144         /**
 145          * @brief Check an URI for LRDD data
 146          *
 147          * this is a replacement for the "lrdd" function in include/network.php.
 148          * It isn't used in this class and has some redundancies in the code.
 149          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 150          *
 151          * @param string $uri Address that should be probed
 152          *
 153          * @return array uri data
 154          */
 155         public static function lrdd($uri) {
 156
 157                 $lrdd = self::xrd($uri);
 158
 159                 if (!$lrdd) {
 160                         $parts = @parse_url($uri);
 161                         if (!$parts)
 162                                 return array();
 163
 164                         $host = $parts["host"];
 165
 166                         $path_parts = explode("/", trim($parts["path"], "/"));
 167
 168                         do {
 169                                 $lrdd = self::xrd($host);
 170                                 $host .= "/".array_shift($path_parts);
 171                         } while (!$lrdd AND (sizeof($path_parts) > 0));
 172                 }
 173
 174                 if (!$lrdd)
 175                         return array();
 176
 177                 foreach ($lrdd AS $key => $link) {
 178                         if ($webfinger)
 179                                 continue;
 180
 181                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
 182                                 continue;
 183
 184                         $path = str_replace('{uri}', urlencode($uri), $link);
 185                         $webfinger = self::webfinger($path);
 186
 187                         if (!$webfinger AND (strstr($uri, "@"))) {
 188                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
 189                                 $webfinger = self::webfinger($path);
 190                         }
 191                 }
 192
 193                 if (!is_array($webfinger["links"]))
 194                         return false;
 195
 196                 $data = array();
 197
 198                 foreach ($webfinger["links"] AS $link)
 199                         $data[] = array("@attributes" => $link);
 200
 201                 if (is_array($webfinger["aliases"]))
 202                         foreach ($webfinger["aliases"] AS $alias)
 203                                 $data[] = array("@attributes" =>
 204                                                         array("rel" => "alias",
 205                                                                 "href" => $alias));
 206
 207                 return $data;
 208         }
 209
 210         /**
 211          * @brief Fetch information (protocol endpoints and user information) about a given uri
 212          *
 213          * @param string $uri Address that should be probed
 214          * @param string $network Test for this specific network
 215          * @param integer $uid User ID for the probe (only used for mails)
 216          * @param boolean $cache Use cached values?
 217          *
 218          * @return array uri data
 219          */
 220         public static function uri($uri, $network = "", $uid = 0, $cache = true) {
 221
 222                 if ($cache) {
 223                         $result = Cache::get("probe_url:".$network.":".$uri);
 224                         if (!is_null($result)) {
 225                                 return $result;
 226                         }
 227                 }
 228
 229                 if ($uid == 0)
 230                         $uid = local_user();
 231
 232                 $data = self::detect($uri, $network, $uid);
 233
 234                 if (!isset($data["url"]))
 235                         $data["url"] = $uri;
 236
 237                 if ($data["photo"] != "")
 238                         $data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
 239                 else
 240                         $data["photo"] = App::get_baseurl().'/images/person-175.jpg';
 241
 242                 if (!isset($data["name"]) OR ($data["name"] == "")) {
 243                         if (isset($data["nick"]))
 244                                 $data["name"] = $data["nick"];
 245
 246                         if ($data["name"] == "")
 247                                 $data["name"] = $data["url"];
 248                 }
 249
 250                 if (!isset($data["nick"]) OR ($data["nick"] == "")) {
 251                         $data["nick"] = strtolower($data["name"]);
 252
 253                         if (strpos($data['nick'], ' '))
 254                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 255                 }
 256
 257                 if (!isset($data["network"]))
 258                         $data["network"] = NETWORK_PHANTOM;
 259
 260                 $data = self::rearrange_data($data);
 261
 262                 // Only store into the cache if the value seems to be valid
 263                 if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
 264                         Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
 265
 266                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 267                         /// The biggest problem is the avatar picture that could have a reduced image size.
 268                         /// It should only be updated if the existing picture isn't existing anymore.
 269                         if (($data['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
 270                                 $data["name"] AND $data["nick"] AND $data["url"] AND $data["addr"] AND $data["poll"])
 271                                 q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 272                                                 `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 273                                         WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 274                                         dbesc($data["name"]),
 275                                         dbesc($data["nick"]),
 276                                         dbesc($data["url"]),
 277                                         dbesc($data["addr"]),
 278                                         dbesc($data["notify"]),
 279                                         dbesc($data["poll"]),
 280                                         dbesc($data["alias"]),
 281                                         dbesc(datetime_convert()),
 282                                         dbesc(normalise_link($data['url']))
 283                         );
 284                 }
 285                 return $data;
 286         }
 287
 288         /**
 289          * @brief Fetch information (protocol endpoints and user information) about a given uri
 290          *
 291          * This function is only called by the "uri" function that adds caching and rearranging of data.
 292          *
 293          * @param string $uri Address that should be probed
 294          * @param string $network Test for this specific network
 295          * @param integer $uid User ID for the probe (only used for mails)
 296          *
 297          * @return array uri data
 298          */
 299         private function detect($uri, $network, $uid) {
 300                 if (strstr($uri, '@')) {
 301                         // If the URI starts with "mailto:" then jump directly to the mail detection
 302                         if (strpos($url,'mailto:') !== false) {
 303                                 $uri = str_replace('mailto:', '', $url);
 304                                 return self::mail($uri, $uid);
 305                         }
 306
 307                         if ($network == NETWORK_MAIL)
 308                                 return self::mail($uri, $uid);
 309
 310                         // Remove "acct:" from the URI
 311                         $uri = str_replace('acct:', '', $uri);
 312
 313                         $host = substr($uri,strpos($uri, '@') + 1);
 314                         $nick = substr($uri,0, strpos($uri, '@'));
 315
 316                         if (strpos($uri, '@twitter.com'))
 317                                 return array("network" => NETWORK_TWITTER);
 318
 319                         $lrdd = self::xrd($host);
 320
 321                         if (!$lrdd)
 322                                 return self::mail($uri, $uid);
 323
 324                         $addr = $uri;
 325                 } else {
 326                         $parts = parse_url($uri);
 327                         if (!isset($parts["scheme"]) OR
 328                                 !isset($parts["host"]) OR
 329                                 !isset($parts["path"]))
 330                                 return false;
 331
 332                         /// @todo: Ports?
 333                         $host = $parts["host"];
 334
 335                         if ($host == 'twitter.com')
 336                                 return array("network" => NETWORK_TWITTER);
 337
 338                         $lrdd = self::xrd($host);
 339
 340                         $path_parts = explode("/", trim($parts["path"], "/"));
 341
 342                         while (!$lrdd AND (sizeof($path_parts) > 1)) {
 343                                 $host .= "/".array_shift($path_parts);
 344                                 $lrdd = self::xrd($host);
 345                         }
 346                         if (!$lrdd)
 347                                 return self::feed($uri);
 348
 349                         $nick = array_pop($path_parts);
 350                         $addr = $nick."@".$host;
 351                 }
 352                 $webfinger = false;
 353
 354                 /// @todo Do we need the prefix "acct:" or "acct://"?
 355
 356                 foreach ($lrdd AS $key => $link) {
 357                         if ($webfinger)
 358                                 continue;
 359
 360                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
 361                                 continue;
 362
 363                         // Try webfinger with the address (user@domain.tld)
 364                         $path = str_replace('{uri}', urlencode($addr), $link);
 365                         $webfinger = self::webfinger($path);
 366
 367                         // Mastodon needs to have it with "acct:"
 368                         if (!$webfinger) {
 369                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 370                                 $webfinger = self::webfinger($path);
 371                         }
 372
 373                         // If webfinger wasn't successful then try it with the URL - possibly in the format https://...
 374                         if (!$webfinger AND ($uri != $addr)) {
 375                                 $path = str_replace('{uri}', urlencode($uri), $link);
 376                                 $webfinger = self::webfinger($path);
 377
 378                                 // Since the detection with the address wasn't successful, we delete it.
 379                                 if ($webfinger) {
 380                                         $nick = "";
 381                                         $addr = "";
 382                                 }
 383                         }
 384
 385                 }
 386                 if (!$webfinger)
 387                         return self::feed($uri);
 388
 389                 $result = false;
 390
 391                 logger("Probing ".$uri, LOGGER_DEBUG);
 392
 393                 if (in_array($network, array("", NETWORK_DFRN)))
 394                         $result = self::dfrn($webfinger);
 395                 if ((!$result AND ($network == "")) OR ($network == NETWORK_DIASPORA))
 396                         $result = self::diaspora($webfinger);
 397                 if ((!$result AND ($network == "")) OR ($network == NETWORK_OSTATUS))
 398                         $result = self::ostatus($webfinger);
 399                 if ((!$result AND ($network == "")) OR ($network == NETWORK_PUMPIO))
 400                         $result = self::pumpio($webfinger);
 401                 if ((!$result AND ($network == "")) OR ($network == NETWORK_FEED))
 402                         $result = self::feed($uri);
 403                 else {
 404                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 405                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 406                         if ((!isset($result["nick"]) OR ($result["nick"] == "") OR (strstr($result["nick"], " "))) AND ($nick != ""))
 407                                 $result["nick"] = $nick;
 408
 409                         if ((!isset($result["addr"]) OR ($result["addr"] == "")) AND ($addr != ""))
 410                                 $result["addr"] = $addr;
 411                 }
 412
 413                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 414
 415                 if (!isset($result["baseurl"]) OR ($result["baseurl"] == "")) {
 416                         $pos = strpos($result["url"], $host);
 417                         if ($pos)
 418                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 419                 }
 420
 421                 return $result;
 422         }
 423
 424         /**
 425          * @brief Perform a webfinger request.
 426          *
 427          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 428          *
 429          * @param string $url Address that should be probed
 430          *
 431          * @return array webfinger data
 432          */
 433         private function webfinger($url) {
 434
 435                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
 436                 $redirects = 0;
 437
 438                 $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 439                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 440                         return false;
 441                 }
 442                 $data = $ret['body'];
 443
 444                 $xrd = parse_xml_string($data, false);
 445
 446                 if (!is_object($xrd)) {
 447                         // If it is not XML, maybe it is JSON
 448                         $webfinger = json_decode($data, true);
 449
 450                         if (!isset($webfinger["links"]))
 451                                 return false;
 452
 453                         return $webfinger;
 454                 }
 455
 456                 $xrd_arr = xml::element_to_array($xrd);
 457                 if (!isset($xrd_arr["xrd"]["link"]))
 458                         return false;
 459
 460                 $webfinger = array();
 461
 462                 if (isset($xrd_arr["xrd"]["subject"]))
 463                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 464
 465                 if (isset($xrd_arr["xrd"]["alias"]))
 466                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 467
 468                 $webfinger["links"] = array();
 469
 470                 foreach ($xrd_arr["xrd"]["link"] AS $value => $data) {
 471                         if (isset($data["@attributes"]))
 472                                 $attributes = $data["@attributes"];
 473                         elseif ($value == "@attributes")
 474                                 $attributes = $data;
 475                         else
 476                                 continue;
 477
 478                         $webfinger["links"][] = $attributes;
 479                 }
 480                 return $webfinger;
 481         }
 482
 483         /**
 484          * @brief Poll the Friendica specific noscrape page.
 485          *
 486          * "noscrape" is a faster alternative to fetch the data from the hcard.
 487          * This functionality was originally created for the directory.
 488          *
 489          * @param string $noscrape Link to the noscrape page
 490          * @param array $data The already fetched data
 491          *
 492          * @return array noscrape data
 493          */
 494         private function poll_noscrape($noscrape, $data) {
 495                 $ret = z_fetch_url($noscrape);
 496                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 497                         return false;
 498                 }
 499                 $content = $ret['body'];
 500                 if (!$content) {
 501                         return false;
 502                 }
 503
 504                 $json = json_decode($content, true);
 505                 if (!is_array($json))
 506                         return false;
 507
 508                 if (isset($json["fn"]))
 509                         $data["name"] = $json["fn"];
 510
 511                 if (isset($json["addr"]))
 512                         $data["addr"] = $json["addr"];
 513
 514                 if (isset($json["nick"]))
 515                         $data["nick"] = $json["nick"];
 516
 517                 if (isset($json["comm"]))
 518                         $data["community"] = $json["comm"];
 519
 520                 if (isset($json["tags"])) {
 521                         $keywords = implode(" ", $json["tags"]);
 522                         if ($keywords != "")
 523                                 $data["keywords"] = $keywords;
 524                 }
 525
 526                 $location = formatted_location($json);
 527                 if ($location)
 528                         $data["location"] = $location;
 529
 530                 if (isset($json["about"]))
 531                         $data["about"] = $json["about"];
 532
 533                 if (isset($json["key"]))
 534                         $data["pubkey"] = $json["key"];
 535
 536                 if (isset($json["photo"]))
 537                         $data["photo"] = $json["photo"];
 538
 539                 if (isset($json["dfrn-request"]))
 540                         $data["request"] = $json["dfrn-request"];
 541
 542                 if (isset($json["dfrn-confirm"]))
 543                         $data["confirm"] = $json["dfrn-confirm"];
 544
 545                 if (isset($json["dfrn-notify"]))
 546                         $data["notify"] = $json["dfrn-notify"];
 547
 548                 if (isset($json["dfrn-poll"]))
 549                         $data["poll"] = $json["dfrn-poll"];
 550
 551                 return $data;
 552         }
 553
 554         /**
 555          * @brief Check for valid DFRN data
 556          *
 557          * @param array $data DFRN data
 558          *
 559          * @return int Number of errors
 560          */
 561         public static function valid_dfrn($data) {
 562                 $errors = 0;
 563                 if(!isset($data['key']))
 564                         $errors ++;
 565                 if(!isset($data['dfrn-request']))
 566                         $errors ++;
 567                 if(!isset($data['dfrn-confirm']))
 568                         $errors ++;
 569                 if(!isset($data['dfrn-notify']))
 570                         $errors ++;
 571                 if(!isset($data['dfrn-poll']))
 572                         $errors ++;
 573                 return $errors;
 574         }
 575
 576         /**
 577          * @brief Fetch data from a DFRN profile page and via "noscrape"
 578          *
 579          * @param string $profile Link to the profile page
 580          *
 581          * @return array profile data
 582          */
 583         public static function profile($profile) {
 584
 585                 $data = array();
 586
 587                 logger("Check profile ".$profile, LOGGER_DEBUG);
 588
 589                 // Fetch data via noscrape - this is faster
 590                 $noscrape = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile);
 591                 $data = self::poll_noscrape($noscrape, $data);
 592
 593                 if (!isset($data["notify"]) OR !isset($data["confirm"]) OR
 594                         !isset($data["request"]) OR !isset($data["poll"]) OR
 595                         !isset($data["poco"]) OR !isset($data["name"]) OR
 596                         !isset($data["photo"]))
 597                         $data = self::poll_hcard($profile, $data, true);
 598
 599                 $prof_data = array();
 600                 $prof_data["addr"] = $data["addr"];
 601                 $prof_data["nick"] = $data["nick"];
 602                 $prof_data["dfrn-request"] = $data["request"];
 603                 $prof_data["dfrn-confirm"] = $data["confirm"];
 604                 $prof_data["dfrn-notify"] = $data["notify"];
 605                 $prof_data["dfrn-poll"] = $data["poll"];
 606                 $prof_data["dfrn-poco"] = $data["poco"];
 607                 $prof_data["photo"] = $data["photo"];
 608                 $prof_data["fn"] = $data["name"];
 609                 $prof_data["key"] = $data["pubkey"];
 610
 611                 logger("Result for profile ".$profile.": ".print_r($prof_data, true), LOGGER_DEBUG);
 612
 613                 return $prof_data;
 614         }
 615
 616         /**
 617          * @brief Check for DFRN contact
 618          *
 619          * @param array $webfinger Webfinger data
 620          *
 621          * @return array DFRN data
 622          */
 623         private function dfrn($webfinger) {
 624
 625                 $hcard = "";
 626                 $data = array();
 627                 foreach ($webfinger["links"] AS $link) {
 628                         if (($link["rel"] == NAMESPACE_DFRN) AND ($link["href"] != ""))
 629                                 $data["network"] = NETWORK_DFRN;
 630                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 631                                 $data["poll"] = $link["href"];
 632                         elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 633                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 634                                 $data["url"] = $link["href"];
 635                         elseif (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 636                                 $hcard = $link["href"];
 637                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 638                                 $data["poco"] = $link["href"];
 639                         elseif (($link["rel"] == "http://webfinger.net/rel/avatar") AND ($link["href"] != ""))
 640                                 $data["photo"] = $link["href"];
 641
 642                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 643                                 $data["baseurl"] = trim($link["href"], '/');
 644                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 645                                 $data["guid"] = $link["href"];
 646                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 647                                 $data["pubkey"] = base64_decode($link["href"]);
 648
 649                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 650                                 if (strstr($data["pubkey"], 'RSA '))
 651                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 652                         }
 653                 }
 654
 655                 if (!isset($data["network"]) OR ($hcard == ""))
 656                         return false;
 657
 658                 // Fetch data via noscrape - this is faster
 659                 $noscrape = str_replace("/hcard/", "/noscrape/", $hcard);
 660                 $data = self::poll_noscrape($noscrape, $data);
 661
 662                 if (isset($data["notify"]) AND isset($data["confirm"]) AND isset($data["request"]) AND
 663                         isset($data["poll"]) AND isset($data["name"]) AND isset($data["photo"]))
 664                         return $data;
 665
 666                 $data = self::poll_hcard($hcard, $data, true);
 667
 668                 return $data;
 669         }
 670
 671         /**
 672          * @brief Poll the hcard page (Diaspora and Friendica specific)
 673          *
 674          * @param string $hcard Link to the hcard page
 675          * @param array $data The already fetched data
 676          * @param boolean $dfrn Poll DFRN specific data
 677          *
 678          * @return array hcard data
 679          */
 680         private function poll_hcard($hcard, $data, $dfrn = false) {
 681                 $ret = z_fetch_url($hcard);
 682                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 683                         return false;
 684                 }
 685                 $content = $ret['body'];
 686                 if (!$content) {
 687                         return false;
 688                 }
 689
 690                 $doc = new DOMDocument();
 691                 if (!@$doc->loadHTML($content))
 692                         return false;
 693
 694                 $xpath = new DomXPath($doc);
 695
 696                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 697                 if (!is_object($vcards))
 698                         return false;
 699
 700                 if ($vcards->length > 0) {
 701                         $vcard = $vcards->item(0);
 702
 703                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 704                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 705                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 706                         if (($search->length > 0) AND ($data["guid"] == ""))
 707                                 $data["guid"] = $search->item(0)->nodeValue;
 708
 709                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 710                         if ($search->length > 0)
 711                                 $data["nick"] = $search->item(0)->nodeValue;
 712
 713                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 714                         if ($search->length > 0)
 715                                 $data["name"] = $search->item(0)->nodeValue;
 716
 717                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
 718                         if ($search->length > 0)
 719                                 $data["searchable"] = $search->item(0)->nodeValue;
 720
 721                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
 722                         if ($search->length > 0) {
 723                                 $data["pubkey"] = $search->item(0)->nodeValue;
 724                                 if (strstr($data["pubkey"], 'RSA '))
 725                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 726                         }
 727
 728                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
 729                         if ($search->length > 0)
 730                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
 731                 }
 732
 733                 $avatar = array();
 734                 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
 735                 foreach ($photos AS $photo) {
 736                         $attr = array();
 737                         foreach ($photo->attributes as $attribute) {
 738                                 $attr[$attribute->name] = trim($attribute->value);
 739                         }
 740
 741                         if (isset($attr["src"]) AND isset($attr["width"])) {
 742                                 $avatar[$attr["width"]] = $attr["src"];
 743                         }
 744
 745                         // We don't have a width. So we just take everything that we got.
 746                         // This is a Hubzilla workaround which doesn't send a width.
 747                         if ((sizeof($avatar) == 0) AND isset($attr["src"])) {
 748                                 $avatar[] = $attr["src"];
 749                         }
 750                 }
 751
 752                 if (sizeof($avatar)) {
 753                         ksort($avatar);
 754                         $data["photo"] = array_pop($avatar);
 755                 }
 756
 757                 if ($dfrn) {
 758                         // Poll DFRN specific data
 759                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
 760                         if ($search->length > 0) {
 761                                 foreach ($search AS $link) {
 762                                         //$data["request"] = $search->item(0)->nodeValue;
 763                                         $attr = array();
 764                                         foreach ($link->attributes as $attribute)
 765                                                 $attr[$attribute->name] = trim($attribute->value);
 766
 767                                         $data[substr($attr["rel"], 5)] = $attr["href"];
 768                                 }
 769                         }
 770
 771                         // Older Friendica versions had used the "uid" field differently than newer versions
 772                         if ($data["nick"] == $data["guid"])
 773                                 unset($data["guid"]);
 774                 }
 775
 776
 777                 return $data;
 778         }
 779
 780         /**
 781          * @brief Check for Diaspora contact
 782          *
 783          * @param array $webfinger Webfinger data
 784          *
 785          * @return array Diaspora data
 786          */
 787         private function diaspora($webfinger) {
 788
 789                 $hcard = "";
 790                 $data = array();
 791                 foreach ($webfinger["links"] AS $link) {
 792                         if (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 793                                 $hcard = $link["href"];
 794                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 795                                 $data["baseurl"] = trim($link["href"], '/');
 796                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 797                                 $data["guid"] = $link["href"];
 798                         elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 799                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 800                                 $data["url"] = $link["href"];
 801                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 802                                 $data["poll"] = $link["href"];
 803                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 804                                 $data["poco"] = $link["href"];
 805                         elseif (($link["rel"] == "salmon") AND ($link["href"] != ""))
 806                                 $data["notify"] = $link["href"];
 807                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 808                                 $data["pubkey"] = base64_decode($link["href"]);
 809
 810                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 811                                 if (strstr($data["pubkey"], 'RSA '))
 812                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 813                         }
 814                 }
 815
 816                 if (!isset($data["url"]) OR ($hcard == ""))
 817                         return false;
 818
 819                 if (is_array($webfinger["aliases"]))
 820                         foreach ($webfinger["aliases"] AS $alias)
 821                                 if (normalise_link($alias) != normalise_link($data["url"]) AND !strstr($alias, "@"))
 822                                         $data["alias"] = $alias;
 823
 824                 // Fetch further information from the hcard
 825                 $data = self::poll_hcard($hcard, $data);
 826
 827                 if (!$data)
 828                         return false;
 829
 830                 if (isset($data["url"]) AND isset($data["guid"]) AND isset($data["baseurl"]) AND
 831                         isset($data["pubkey"]) AND ($hcard != "")) {
 832                         $data["network"] = NETWORK_DIASPORA;
 833
 834                         // The Diaspora handle must always be lowercase
 835                         $data["addr"] = strtolower($data["addr"]);
 836
 837                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
 838                         $data["notify"] = $data["baseurl"]."/receive/users/".$data["guid"];
 839                         $data["batch"] = $data["baseurl"]."/receive/public";
 840                 } else
 841                         return false;
 842
 843                 return $data;
 844         }
 845
 846         /**
 847          * @brief Check for OStatus contact
 848          *
 849          * @param array $webfinger Webfinger data
 850          *
 851          * @return array OStatus data
 852          */
 853         private function ostatus($webfinger) {
 854
 855                 $data = array();
 856                 if (is_array($webfinger["aliases"]))
 857                         foreach($webfinger["aliases"] AS $alias)
 858                                 if (strstr($alias, "@"))
 859                                         $data["addr"] = str_replace('acct:', '', $alias);
 860
 861                 if (is_string($webfinger["subject"]) AND strstr($webfinger["subject"], "@"))
 862                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
 863
 864                 $pubkey = "";
 865                 foreach ($webfinger["links"] AS $link) {
 866                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 867                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 868                                 $data["url"] = $link["href"];
 869                         elseif (($link["rel"] == "salmon") AND ($link["href"] != ""))
 870                                 $data["notify"] = $link["href"];
 871                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 872                                 $data["poll"] = $link["href"];
 873                         elseif (($link["rel"] == "magic-public-key") AND ($link["href"] != "")) {
 874                                 $pubkey = $link["href"];
 875
 876                                 if (substr($pubkey, 0, 5) === 'data:') {
 877                                         if (strstr($pubkey, ','))
 878                                                 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
 879                                         else
 880                                                 $pubkey = substr($pubkey, 5);
 881                                 } elseif (normalise_link($pubkey) == 'http://')
 882                                         $pubkey = fetch_url($pubkey);
 883
 884                                 $key = explode(".", $pubkey);
 885
 886                                 if (sizeof($key) >= 3) {
 887                                         $m = base64url_decode($key[1]);
 888                                         $e = base64url_decode($key[2]);
 889                                         $data["pubkey"] = metopem($m,$e);
 890                                 }
 891
 892                         }
 893                 }
 894
 895                 if (isset($data["notify"]) AND isset($data["pubkey"]) AND
 896                         isset($data["poll"]) AND isset($data["url"])) {
 897                         $data["network"] = NETWORK_OSTATUS;
 898                 } else
 899                         return false;
 900
 901                 // Fetch all additional data from the feed
 902                 $feed = fetch_url($data["poll"]);
 903                 $feed_data = feed_import($feed,$dummy1,$dummy2, $dummy3, true);
 904                 if (!$feed_data)
 905                         return false;
 906
 907                 if ($feed_data["header"]["author-name"] != "")
 908                         $data["name"] = $feed_data["header"]["author-name"];
 909
 910                 if ($feed_data["header"]["author-nick"] != "")
 911                         $data["nick"] = $feed_data["header"]["author-nick"];
 912
 913                 if ($feed_data["header"]["author-avatar"] != "")
 914                         $data["photo"] = $feed_data["header"]["author-avatar"];
 915
 916                 if ($feed_data["header"]["author-id"] != "")
 917                         $data["alias"] = $feed_data["header"]["author-id"];
 918
 919                 if ($feed_data["header"]["author-location"] != "")
 920                         $data["location"] = $feed_data["header"]["author-location"];
 921
 922                 if ($feed_data["header"]["author-about"] != "")
 923                         $data["about"] = $feed_data["header"]["author-about"];
 924
 925                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
 926                 // So we take the value that we just fetched, although the other one worked as well
 927                 if ($feed_data["header"]["author-link"] != "")
 928                         $data["url"] = $feed_data["header"]["author-link"];
 929
 930                 /// @todo Fetch location and "about" from the feed as well
 931                 return $data;
 932         }
 933
 934         /**
 935          * @brief Fetch data from a pump.io profile page
 936          *
 937          * @param string $profile Link to the profile page
 938          *
 939          * @return array profile data
 940          */
 941         private function pumpio_profile_data($profile) {
 942
 943                 $doc = new DOMDocument();
 944                 if (!@$doc->loadHTMLFile($profile))
 945                         return false;
 946
 947                 $xpath = new DomXPath($doc);
 948
 949                 $data = array();
 950
 951                 // This is ugly - but pump.io doesn't seem to know a better way for it
 952                 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
 953                 $pos = strpos($data["name"], chr(10));
 954                 if ($pos)
 955                         $data["name"] = trim(substr($data["name"], 0, $pos));
 956
 957                 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
 958                 if ($avatar)
 959                         foreach ($avatar->attributes as $attribute)
 960                                 if ($attribute->name == "src")
 961                                         $data["photo"] = trim($attribute->value);
 962
 963                 $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
 964                 $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
 965
 966                 return $data;
 967         }
 968
 969         /**
 970          * @brief Check for pump.io contact
 971          *
 972          * @param array $webfinger Webfinger data
 973          *
 974          * @return array pump.io data
 975          */
 976         private function pumpio($webfinger) {
 977
 978                 $data = array();
 979                 foreach ($webfinger["links"] AS $link) {
 980                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 981                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 982                                 $data["url"] = $link["href"];
 983                         elseif (($link["rel"] == "activity-inbox") AND ($link["href"] != ""))
 984                                 $data["notify"] = $link["href"];
 985                         elseif (($link["rel"] == "activity-outbox") AND ($link["href"] != ""))
 986                                 $data["poll"] = $link["href"];
 987                         elseif (($link["rel"] == "dialback") AND ($link["href"] != ""))
 988                                 $data["dialback"] = $link["href"];
 989                 }
 990                 if (isset($data["poll"]) AND isset($data["notify"]) AND
 991                         isset($data["dialback"]) AND isset($data["url"])) {
 992
 993                         // by now we use these fields only for the network type detection
 994                         // So we unset all data that isn't used at the moment
 995                         unset($data["dialback"]);
 996
 997                         $data["network"] = NETWORK_PUMPIO;
 998                 } else
 999                         return false;
1000
1001                 $profile_data = self::pumpio_profile_data($data["url"]);
1002
1003                 if (!$profile_data)
1004                         return false;
1005
1006                 $data = array_merge($data, $profile_data);
1007
1008                 return $data;
1009         }
1010
1011         /**
1012          * @brief Check page for feed link
1013          *
1014          * @param string $url Page link
1015          *
1016          * @return string feed link
1017          */
1018         private function get_feed_link($url) {
1019                 $doc = new DOMDocument();
1020
1021                 if (!@$doc->loadHTMLFile($url))
1022                         return false;
1023
1024                 $xpath = new DomXPath($doc);
1025
1026                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1027                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1028                 if (!is_object($feeds))
1029                         return false;
1030
1031                 if ($feeds->length == 0)
1032                         return false;
1033
1034                 $feed_url = "";
1035
1036                 foreach ($feeds AS $feed) {
1037                         $attr = array();
1038                         foreach ($feed->attributes as $attribute)
1039                         $attr[$attribute->name] = trim($attribute->value);
1040
1041                         if ($feed_url == "")
1042                                 $feed_url = $attr["href"];
1043                 }
1044
1045                 return $feed_url;
1046         }
1047
1048         /**
1049          * @brief Check for feed contact
1050          *
1051          * @param string $url Profile link
1052          * @param boolean $probe Do a probe if the page contains a feed link
1053          *
1054          * @return array feed data
1055          */
1056         private function feed($url, $probe = true) {
1057                 $ret = z_fetch_url($url);
1058                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1059                         return false;
1060                 }
1061                 $feed = $ret['body'];
1062                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1063
1064                 if (!$feed_data) {
1065                         if (!$probe)
1066                                 return false;
1067
1068                         $feed_url = self::get_feed_link($url);
1069
1070                         if (!$feed_url)
1071                                 return false;
1072
1073                         return self::feed($feed_url, false);
1074                 }
1075
1076                 if ($feed_data["header"]["author-name"] != "")
1077                         $data["name"] = $feed_data["header"]["author-name"];
1078
1079                 if ($feed_data["header"]["author-nick"] != "")
1080                         $data["nick"] = $feed_data["header"]["author-nick"];
1081
1082                 if ($feed_data["header"]["author-avatar"] != "")
1083                         $data["photo"] = $feed_data["header"]["author-avatar"];
1084
1085                 if ($feed_data["header"]["author-id"] != "")
1086                         $data["alias"] = $feed_data["header"]["author-id"];
1087
1088                 $data["url"] = $url;
1089                 $data["poll"] = $url;
1090
1091                 if ($feed_data["header"]["author-link"] != "")
1092                         $data["baseurl"] = $feed_data["header"]["author-link"];
1093                 else
1094                         $data["baseurl"] = $data["url"];
1095
1096                 $data["network"] = NETWORK_FEED;
1097
1098                 return $data;
1099         }
1100
1101         /**
1102          * @brief Check for mail contact
1103          *
1104          * @param string $uri Profile link
1105          * @param integer $uid User ID
1106          *
1107          * @return array mail data
1108          */
1109         private function mail($uri, $uid) {
1110
1111                 if (!validate_email($uri))
1112                         return false;
1113
1114                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1115
1116                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1117
1118                 if (dbm::is_result($x) && dbm::is_result($r)) {
1119                         $mailbox = construct_mailbox_name($r[0]);
1120                         $password = '';
1121                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password,$x[0]['prvkey']);
1122                         $mbox = email_connect($mailbox,$r[0]['user'], $password);
1123                         if(!mbox)
1124                                 return false;
1125                 }
1126
1127                 $msgs = email_poll($mbox, $uri);
1128                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1129
1130                 if (!count($msgs))
1131                         return false;
1132
1133                 $data = array();
1134
1135                 $data["addr"] = $uri;
1136                 $data["network"] = NETWORK_MAIL;
1137                 $data["name"] = substr($uri, 0, strpos($uri,'@'));
1138                 $data["nick"] = $data["name"];
1139                 $data["photo"] = avatar_img($uri);
1140
1141                 $phost = substr($uri, strpos($uri,'@') + 1);
1142                 $data["url"] = 'http://'.$phost."/".$data["nick"];
1143                 $data["notify"] = 'smtp '.random_string();
1144                 $data["poll"] = 'email '.random_string();
1145
1146                 $x = email_msg_meta($mbox, $msgs[0]);
1147                 if(stristr($x[0]->from, $uri))
1148                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1149                 elseif(stristr($x[0]->to, $uri))
1150                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1151                 if(isset($adr)) {
1152                         foreach($adr as $feadr) {
1153                                 if((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1154                                         &&(strcasecmp($feadr->host, $phost) == 0)
1155                                         && (strlen($feadr->personal))) {
1156
1157                                         $personal = imap_mime_header_decode($feadr->personal);
1158                                         $data["name"] = "";
1159                                         foreach($personal as $perspart)
1160                                                 if ($perspart->charset != "default")
1161                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1162                                                 else
1163                                                         $data["name"] .= $perspart->text;
1164
1165                                         $data["name"] = notags($data["name"]);
1166                                 }
1167                         }
1168                 }
1169                 imap_close($mbox);
1170
1171                 return $data;
1172         }
1173 }
1174 ?>