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