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