include/Probe.php

   1 <?php
   2 /**
   3  * @file include/Probe.php
   4  * @brief Functions for probing URL
   5  *
   6  */
   7
   8 use Friendica\App;
   9 use Friendica\Core\Config;
  10 use Friendica\Core\PConfig;
  11
  12 require_once("include/feed.php");
  13 require_once('include/email.php');
  14 require_once('include/network.php');
  15
  16 /**
  17  * @brief This class contain functions for probing URL
  18  *
  19  */
  20 class Probe {
  21
  22         private static $baseurl;
  23
  24         /**
  25          * @brief Rearrange the array so that it always has the same order
  26          *
  27          * @param array $data Unordered data
  28          *
  29          * @return array Ordered data
  30          */
  31         private function rearrange_data($data) {
  32                 $fields = array("name", "nick", "guid", "url", "addr", "alias",
  33                                 "photo", "community", "keywords", "location", "about",
  34                                 "batch", "notify", "poll", "request", "confirm", "poco",
  35                                 "priority", "network", "pubkey", "baseurl");
  36
  37                 $newdata = array();
  38                 foreach ($fields AS $field)
  39                         if (isset($data[$field]))
  40                                 $newdata[$field] = $data[$field];
  41                         else
  42                                 $newdata[$field] = "";
  43
  44                 // We don't use the "priority" field anymore and replace it with a dummy.
  45                 $newdata["priority"] = 0;
  46
  47                 return $newdata;
  48         }
  49
  50         /**
  51          * @brief Probes for XRD data
  52          *
  53          * @return array
  54          *      'lrdd' => Link to LRDD endpoint
  55          *      'lrdd-xml' => Link to LRDD endpoint in XML format
  56          *      'lrdd-json' => Link to LRDD endpoint in JSON format
  57          */
  58         private function xrd($host) {
  59
  60                 // Reset the static variable
  61                 self::$baseurl = '';
  62
  63                 $ssl_url = "https://".$host."/.well-known/host-meta";
  64                 $url = "http://".$host."/.well-known/host-meta";
  65
  66                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
  67                 $redirects = 0;
  68
  69                 $ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
  70                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
  71                         return false;
  72                 }
  73                 $xml = $ret['body'];
  74
  75                 $xrd = parse_xml_string($xml, false);
  76
  77                 if (!is_object($xrd)) {
  78                         $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
  79                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
  80                                 return false;
  81                         }
  82                         $xml = $ret['body'];
  83                         $xrd = parse_xml_string($xml, false);
  84                 }
  85                 if (!is_object($xrd))
  86                         return false;
  87
  88                 $links = xml::element_to_array($xrd);
  89                 if (!isset($links["xrd"]["link"]))
  90                         return false;
  91
  92                 $xrd_data = array();
  93
  94                 foreach ($links["xrd"]["link"] AS $value => $link) {
  95                         if (isset($link["@attributes"]))
  96                                 $attributes = $link["@attributes"];
  97                         elseif ($value == "@attributes")
  98                                 $attributes = $link;
  99                         else
 100                                 continue;
 101
 102                         if (($attributes["rel"] == "lrdd") AND
 103                                 ($attributes["type"] == "application/xrd+xml"))
 104                                 $xrd_data["lrdd-xml"] = $attributes["template"];
 105                         elseif (($attributes["rel"] == "lrdd") AND
 106                                 ($attributes["type"] == "application/json"))
 107                                 $xrd_data["lrdd-json"] = $attributes["template"];
 108                         elseif ($attributes["rel"] == "lrdd")
 109                                 $xrd_data["lrdd"] = $attributes["template"];
 110                 }
 111
 112                 self::$baseurl = "http://".$host;
 113
 114                 return $xrd_data;
 115         }
 116
 117         /**
 118          * @brief Perform Webfinger lookup and return DFRN data
 119          *
 120          * Given an email style address, perform webfinger lookup and
 121          * return the resulting DFRN profile URL, or if no DFRN profile URL
 122          * is located, returns an OStatus subscription template (prefixed
 123          * with the string 'stat:' to identify it as on OStatus template).
 124          * If this isn't an email style address just return $webbie.
 125          * Return an empty string if email-style addresses but webfinger fails,
 126          * or if the resultant personal XRD doesn't contain a supported
 127          * subscription/friend-request attribute.
 128          *
 129          * amended 7/9/2011 to return an hcard which could save potentially loading
 130          * a lengthy content page to scrape dfrn attributes
 131          *
 132          * @param string $webbie Address that should be probed
 133          * @param string $hcard Link to the hcard - is returned by reference
 134          *
 135          * @return string profile link
 136          */
 137
 138         public static function webfinger_dfrn($webbie, &$hcard) {
 139
 140                 $profile_link = '';
 141
 142                 $links = self::lrdd($webbie);
 143                 logger('webfinger_dfrn: '.$webbie.':'.print_r($links,true), LOGGER_DATA);
 144                 if (count($links)) {
 145                         foreach ($links as $link) {
 146                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN)
 147                                         $profile_link = $link['@attributes']['href'];
 148                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) AND ($profile_link == ""))
 149                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 150                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 151                                         $hcard = $link['@attributes']['href'];
 152                         }
 153                 }
 154                 return $profile_link;
 155         }
 156
 157         /**
 158          * @brief Check an URI for LRDD data
 159          *
 160          * this is a replacement for the "lrdd" function in include/network.php.
 161          * It isn't used in this class and has some redundancies in the code.
 162          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 163          *
 164          * @param string $uri Address that should be probed
 165          *
 166          * @return array uri data
 167          */
 168         public static function lrdd($uri) {
 169
 170                 $lrdd = self::xrd($uri);
 171
 172                 if (!$lrdd) {
 173                         $parts = @parse_url($uri);
 174                         if (!$parts)
 175                                 return array();
 176
 177                         $host = $parts["host"];
 178
 179                         $path_parts = explode("/", trim($parts["path"], "/"));
 180
 181                         $nick = array_pop($path_parts);
 182
 183                         do {
 184                                 $lrdd = self::xrd($host);
 185                                 $host .= "/".array_shift($path_parts);
 186                         } while (!$lrdd AND (sizeof($path_parts) > 0));
 187                 }
 188
 189                 if (!$lrdd)
 190                         return array();
 191
 192                 foreach ($lrdd AS $key => $link) {
 193                         if ($webfinger)
 194                                 continue;
 195
 196                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
 197                                 continue;
 198
 199                         $path = str_replace('{uri}', urlencode($uri), $link);
 200                         $webfinger = self::webfinger($path);
 201
 202                         if (!$webfinger AND (strstr($uri, "@"))) {
 203                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
 204                                 $webfinger = self::webfinger($path);
 205                         }
 206
 207                         // Special treatment for Mastodon
 208                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 209                         // But the webfinger for this format fails.
 210                         if (!$webfinger AND isset($nick)) {
 211                                 // Mastodon uses a "@" as prefix for usernames in their url format
 212                                 $nick = ltrim($nick, '@');
 213
 214                                 $addr = $nick."@".$host;
 215
 216                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 217                                 $webfinger = self::webfinger($path);
 218                         }
 219                 }
 220
 221                 if (!is_array($webfinger["links"]))
 222                         return false;
 223
 224                 $data = array();
 225
 226                 foreach ($webfinger["links"] AS $link)
 227                         $data[] = array("@attributes" => $link);
 228
 229                 if (is_array($webfinger["aliases"]))
 230                         foreach ($webfinger["aliases"] AS $alias)
 231                                 $data[] = array("@attributes" =>
 232                                                         array("rel" => "alias",
 233                                                                 "href" => $alias));
 234
 235                 return $data;
 236         }
 237
 238         /**
 239          * @brief Fetch information (protocol endpoints and user information) about a given uri
 240          *
 241          * @param string $uri Address that should be probed
 242          * @param string $network Test for this specific network
 243          * @param integer $uid User ID for the probe (only used for mails)
 244          * @param boolean $cache Use cached values?
 245          *
 246          * @return array uri data
 247          */
 248         public static function uri($uri, $network = "", $uid = 0, $cache = true) {
 249
 250                 if ($cache) {
 251                         $result = Cache::get("probe_url:".$network.":".$uri);
 252                         if (!is_null($result)) {
 253                                 return $result;
 254                         }
 255                 }
 256
 257                 if ($uid == 0)
 258                         $uid = local_user();
 259
 260                 $data = self::detect($uri, $network, $uid);
 261
 262                 if (!isset($data["url"]))
 263                         $data["url"] = $uri;
 264
 265                 if ($data["photo"] != "")
 266                         $data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
 267                 else
 268                         $data["photo"] = App::get_baseurl().'/images/person-175.jpg';
 269
 270                 if (!isset($data["name"]) OR ($data["name"] == "")) {
 271                         if (isset($data["nick"]))
 272                                 $data["name"] = $data["nick"];
 273
 274                         if ($data["name"] == "")
 275                                 $data["name"] = $data["url"];
 276                 }
 277
 278                 if (!isset($data["nick"]) OR ($data["nick"] == "")) {
 279                         $data["nick"] = strtolower($data["name"]);
 280
 281                         if (strpos($data['nick'], ' '))
 282                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 283                 }
 284
 285                 if (self::$baseurl != "") {
 286                         $data["baseurl"] = self::$baseurl;
 287                 }
 288
 289                 if (!isset($data["network"])) {
 290                         $data["network"] = NETWORK_PHANTOM;
 291                 }
 292
 293                 $data = self::rearrange_data($data);
 294
 295                 // Only store into the cache if the value seems to be valid
 296                 if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
 297                         Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
 298
 299                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 300                         /// The biggest problem is the avatar picture that could have a reduced image size.
 301                         /// It should only be updated if the existing picture isn't existing anymore.
 302                         if (($data['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
 303                                 $data["name"] AND $data["nick"] AND $data["url"] AND $data["addr"] AND $data["poll"])
 304                                 q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 305                                                 `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 306                                         WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 307                                         dbesc($data["name"]),
 308                                         dbesc($data["nick"]),
 309                                         dbesc($data["url"]),
 310                                         dbesc($data["addr"]),
 311                                         dbesc($data["notify"]),
 312                                         dbesc($data["poll"]),
 313                                         dbesc($data["alias"]),
 314                                         dbesc(datetime_convert()),
 315                                         dbesc(normalise_link($data['url']))
 316                         );
 317                 }
 318
 319                 return $data;
 320         }
 321
 322         /**
 323          * @brief Fetch information (protocol endpoints and user information) about a given uri
 324          *
 325          * This function is only called by the "uri" function that adds caching and rearranging of data.
 326          *
 327          * @param string $uri Address that should be probed
 328          * @param string $network Test for this specific network
 329          * @param integer $uid User ID for the probe (only used for mails)
 330          *
 331          * @return array uri data
 332          */
 333         private function detect($uri, $network, $uid) {
 334                 $parts = parse_url($uri);
 335
 336                 if (isset($parts["scheme"]) AND isset($parts["host"]) AND isset($parts["path"])) {
 337
 338                         /// @todo: Ports?
 339                         $host = $parts["host"];
 340
 341                         if ($host == 'twitter.com') {
 342                                 return array("network" => NETWORK_TWITTER);
 343                         }
 344                         $lrdd = self::xrd($host);
 345
 346                         $path_parts = explode("/", trim($parts["path"], "/"));
 347
 348                         while (!$lrdd AND (sizeof($path_parts) > 1)) {
 349                                 $host .= "/".array_shift($path_parts);
 350                                 $lrdd = self::xrd($host);
 351                         }
 352                         if (!$lrdd) {
 353                                 return self::feed($uri);
 354                         }
 355                         $nick = array_pop($path_parts);
 356
 357                         // Mastodon uses a "@" as prefix for usernames in their url format
 358                         $nick = ltrim($nick, '@');
 359
 360                         $addr = $nick."@".$host;
 361                 } elseif (strstr($uri, '@')) {
 362                         // If the URI starts with "mailto:" then jump directly to the mail detection
 363                         if (strpos($url,'mailto:') !== false) {
 364                                 $uri = str_replace('mailto:', '', $url);
 365                                 return self::mail($uri, $uid);
 366                         }
 367
 368                         if ($network == NETWORK_MAIL) {
 369                                 return self::mail($uri, $uid);
 370                         }
 371                         // Remove "acct:" from the URI
 372                         $uri = str_replace('acct:', '', $uri);
 373
 374                         $host = substr($uri,strpos($uri, '@') + 1);
 375                         $nick = substr($uri,0, strpos($uri, '@'));
 376
 377                         if (strpos($uri, '@twitter.com')) {
 378                                 return array("network" => NETWORK_TWITTER);
 379                         }
 380                         $lrdd = self::xrd($host);
 381
 382                         if (!$lrdd) {
 383                                 return self::mail($uri, $uid);
 384                         }
 385                         $addr = $uri;
 386                 } else {
 387                         return false;
 388                 }
 389
 390                 $webfinger = false;
 391
 392                 /// @todo Do we need the prefix "acct:" or "acct://"?
 393
 394                 foreach ($lrdd AS $key => $link) {
 395                         if ($webfinger) {
 396                                 continue;
 397                         }
 398                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
 399                                 continue;
 400                         }
 401                         // At first try it with the given uri
 402                         $path = str_replace('{uri}', urlencode($uri), $link);
 403                         $webfinger = self::webfinger($path);
 404
 405                         // We cannot be sure that the detected address was correct, so we don't use the values
 406                         if ($webfinger AND ($uri != $addr)) {
 407                                 $nick = "";
 408                                 $addr = "";
 409                         }
 410
 411                         // Try webfinger with the address (user@domain.tld)
 412                         if (!$webfinger) {
 413                                 $path = str_replace('{uri}', urlencode($addr), $link);
 414                                 $webfinger = self::webfinger($path);
 415                         }
 416
 417                         // Mastodon needs to have it with "acct:"
 418                         if (!$webfinger) {
 419                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 420                                 $webfinger = self::webfinger($path);
 421                         }
 422                 }
 423                 if (!$webfinger) {
 424                         return self::feed($uri);
 425                 }
 426
 427                 $result = false;
 428
 429                 logger("Probing ".$uri, LOGGER_DEBUG);
 430
 431                 if (in_array($network, array("", NETWORK_DFRN)))
 432                         $result = self::dfrn($webfinger);
 433                 if ((!$result AND ($network == "")) OR ($network == NETWORK_DIASPORA))
 434                         $result = self::diaspora($webfinger);
 435                 if ((!$result AND ($network == "")) OR ($network == NETWORK_OSTATUS))
 436                         $result = self::ostatus($webfinger);
 437                 if ((!$result AND ($network == "")) OR ($network == NETWORK_PUMPIO))
 438                         $result = self::pumpio($webfinger);
 439                 if ((!$result AND ($network == "")) OR ($network == NETWORK_FEED))
 440                         $result = self::feed($uri);
 441                 else {
 442                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 443                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 444                         if ((!isset($result["nick"]) OR ($result["nick"] == "") OR (strstr($result["nick"], " "))) AND ($nick != ""))
 445                                 $result["nick"] = $nick;
 446
 447                         if ((!isset($result["addr"]) OR ($result["addr"] == "")) AND ($addr != ""))
 448                                 $result["addr"] = $addr;
 449                 }
 450
 451                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 452
 453                 if (!isset($result["baseurl"]) OR ($result["baseurl"] == "")) {
 454                         $pos = strpos($result["url"], $host);
 455                         if ($pos)
 456                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 457                 }
 458
 459                 return $result;
 460         }
 461
 462         /**
 463          * @brief Perform a webfinger request.
 464          *
 465          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 466          *
 467          * @param string $url Address that should be probed
 468          *
 469          * @return array webfinger data
 470          */
 471         private function webfinger($url) {
 472
 473                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
 474                 $redirects = 0;
 475
 476                 $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 477                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 478                         return false;
 479                 }
 480                 $data = $ret['body'];
 481
 482                 $xrd = parse_xml_string($data, false);
 483
 484                 if (!is_object($xrd)) {
 485                         // If it is not XML, maybe it is JSON
 486                         $webfinger = json_decode($data, true);
 487
 488                         if (!isset($webfinger["links"]))
 489                                 return false;
 490
 491                         return $webfinger;
 492                 }
 493
 494                 $xrd_arr = xml::element_to_array($xrd);
 495                 if (!isset($xrd_arr["xrd"]["link"]))
 496                         return false;
 497
 498                 $webfinger = array();
 499
 500                 if (isset($xrd_arr["xrd"]["subject"]))
 501                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 502
 503                 if (isset($xrd_arr["xrd"]["alias"]))
 504                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 505
 506                 $webfinger["links"] = array();
 507
 508                 foreach ($xrd_arr["xrd"]["link"] AS $value => $data) {
 509                         if (isset($data["@attributes"]))
 510                                 $attributes = $data["@attributes"];
 511                         elseif ($value == "@attributes")
 512                                 $attributes = $data;
 513                         else
 514                                 continue;
 515
 516                         $webfinger["links"][] = $attributes;
 517                 }
 518                 return $webfinger;
 519         }
 520
 521         /**
 522          * @brief Poll the Friendica specific noscrape page.
 523          *
 524          * "noscrape" is a faster alternative to fetch the data from the hcard.
 525          * This functionality was originally created for the directory.
 526          *
 527          * @param string $noscrape Link to the noscrape page
 528          * @param array $data The already fetched data
 529          *
 530          * @return array noscrape data
 531          */
 532         private function poll_noscrape($noscrape, $data) {
 533                 $ret = z_fetch_url($noscrape);
 534                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 535                         return false;
 536                 }
 537                 $content = $ret['body'];
 538                 if (!$content) {
 539                         return false;
 540                 }
 541
 542                 $json = json_decode($content, true);
 543                 if (!is_array($json))
 544                         return false;
 545
 546                 if (isset($json["fn"]))
 547                         $data["name"] = $json["fn"];
 548
 549                 if (isset($json["addr"]))
 550                         $data["addr"] = $json["addr"];
 551
 552                 if (isset($json["nick"]))
 553                         $data["nick"] = $json["nick"];
 554
 555                 if (isset($json["comm"]))
 556                         $data["community"] = $json["comm"];
 557
 558                 if (isset($json["tags"])) {
 559                         $keywords = implode(" ", $json["tags"]);
 560                         if ($keywords != "")
 561                                 $data["keywords"] = $keywords;
 562                 }
 563
 564                 $location = formatted_location($json);
 565                 if ($location)
 566                         $data["location"] = $location;
 567
 568                 if (isset($json["about"]))
 569                         $data["about"] = $json["about"];
 570
 571                 if (isset($json["key"]))
 572                         $data["pubkey"] = $json["key"];
 573
 574                 if (isset($json["photo"]))
 575                         $data["photo"] = $json["photo"];
 576
 577                 if (isset($json["dfrn-request"]))
 578                         $data["request"] = $json["dfrn-request"];
 579
 580                 if (isset($json["dfrn-confirm"]))
 581                         $data["confirm"] = $json["dfrn-confirm"];
 582
 583                 if (isset($json["dfrn-notify"]))
 584                         $data["notify"] = $json["dfrn-notify"];
 585
 586                 if (isset($json["dfrn-poll"]))
 587                         $data["poll"] = $json["dfrn-poll"];
 588
 589                 return $data;
 590         }
 591
 592         /**
 593          * @brief Check for valid DFRN data
 594          *
 595          * @param array $data DFRN data
 596          *
 597          * @return int Number of errors
 598          */
 599         public static function valid_dfrn($data) {
 600                 $errors = 0;
 601                 if(!isset($data['key']))
 602                         $errors ++;
 603                 if(!isset($data['dfrn-request']))
 604                         $errors ++;
 605                 if(!isset($data['dfrn-confirm']))
 606                         $errors ++;
 607                 if(!isset($data['dfrn-notify']))
 608                         $errors ++;
 609                 if(!isset($data['dfrn-poll']))
 610                         $errors ++;
 611                 return $errors;
 612         }
 613
 614         /**
 615          * @brief Fetch data from a DFRN profile page and via "noscrape"
 616          *
 617          * @param string $profile Link to the profile page
 618          *
 619          * @return array profile data
 620          */
 621         public static function profile($profile) {
 622
 623                 $data = array();
 624
 625                 logger("Check profile ".$profile, LOGGER_DEBUG);
 626
 627                 // Fetch data via noscrape - this is faster
 628                 $noscrape = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile);
 629                 $data = self::poll_noscrape($noscrape, $data);
 630
 631                 if (!isset($data["notify"]) OR !isset($data["confirm"]) OR
 632                         !isset($data["request"]) OR !isset($data["poll"]) OR
 633                         !isset($data["poco"]) OR !isset($data["name"]) OR
 634                         !isset($data["photo"]))
 635                         $data = self::poll_hcard($profile, $data, true);
 636
 637                 $prof_data = array();
 638                 $prof_data["addr"] = $data["addr"];
 639                 $prof_data["nick"] = $data["nick"];
 640                 $prof_data["dfrn-request"] = $data["request"];
 641                 $prof_data["dfrn-confirm"] = $data["confirm"];
 642                 $prof_data["dfrn-notify"] = $data["notify"];
 643                 $prof_data["dfrn-poll"] = $data["poll"];
 644                 $prof_data["dfrn-poco"] = $data["poco"];
 645                 $prof_data["photo"] = $data["photo"];
 646                 $prof_data["fn"] = $data["name"];
 647                 $prof_data["key"] = $data["pubkey"];
 648
 649                 logger("Result for profile ".$profile.": ".print_r($prof_data, true), LOGGER_DEBUG);
 650
 651                 return $prof_data;
 652         }
 653
 654         /**
 655          * @brief Check for DFRN contact
 656          *
 657          * @param array $webfinger Webfinger data
 658          *
 659          * @return array DFRN data
 660          */
 661         private function dfrn($webfinger) {
 662
 663                 $hcard = "";
 664                 $data = array();
 665                 foreach ($webfinger["links"] AS $link) {
 666                         if (($link["rel"] == NAMESPACE_DFRN) AND ($link["href"] != ""))
 667                                 $data["network"] = NETWORK_DFRN;
 668                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 669                                 $data["poll"] = $link["href"];
 670                         elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 671                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 672                                 $data["url"] = $link["href"];
 673                         elseif (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 674                                 $hcard = $link["href"];
 675                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 676                                 $data["poco"] = $link["href"];
 677                         elseif (($link["rel"] == "http://webfinger.net/rel/avatar") AND ($link["href"] != ""))
 678                                 $data["photo"] = $link["href"];
 679
 680                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 681                                 $data["baseurl"] = trim($link["href"], '/');
 682                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 683                                 $data["guid"] = $link["href"];
 684                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 685                                 $data["pubkey"] = base64_decode($link["href"]);
 686
 687                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 688                                 if (strstr($data["pubkey"], 'RSA '))
 689                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 690                         }
 691                 }
 692
 693                 if (!isset($data["network"]) OR ($hcard == ""))
 694                         return false;
 695
 696                 // Fetch data via noscrape - this is faster
 697                 $noscrape = str_replace("/hcard/", "/noscrape/", $hcard);
 698                 $data = self::poll_noscrape($noscrape, $data);
 699
 700                 if (isset($data["notify"]) AND isset($data["confirm"]) AND isset($data["request"]) AND
 701                         isset($data["poll"]) AND isset($data["name"]) AND isset($data["photo"]))
 702                         return $data;
 703
 704                 $data = self::poll_hcard($hcard, $data, true);
 705
 706                 return $data;
 707         }
 708
 709         /**
 710          * @brief Poll the hcard page (Diaspora and Friendica specific)
 711          *
 712          * @param string $hcard Link to the hcard page
 713          * @param array $data The already fetched data
 714          * @param boolean $dfrn Poll DFRN specific data
 715          *
 716          * @return array hcard data
 717          */
 718         private function poll_hcard($hcard, $data, $dfrn = false) {
 719                 $ret = z_fetch_url($hcard);
 720                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 721                         return false;
 722                 }
 723                 $content = $ret['body'];
 724                 if (!$content) {
 725                         return false;
 726                 }
 727
 728                 $doc = new DOMDocument();
 729                 if (!@$doc->loadHTML($content))
 730                         return false;
 731
 732                 $xpath = new DomXPath($doc);
 733
 734                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 735                 if (!is_object($vcards))
 736                         return false;
 737
 738                 if ($vcards->length > 0) {
 739                         $vcard = $vcards->item(0);
 740
 741                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 742                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 743                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 744                         if (($search->length > 0) AND ($data["guid"] == ""))
 745                                 $data["guid"] = $search->item(0)->nodeValue;
 746
 747                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 748                         if ($search->length > 0)
 749                                 $data["nick"] = $search->item(0)->nodeValue;
 750
 751                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 752                         if ($search->length > 0)
 753                                 $data["name"] = $search->item(0)->nodeValue;
 754
 755                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
 756                         if ($search->length > 0)
 757                                 $data["searchable"] = $search->item(0)->nodeValue;
 758
 759                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
 760                         if ($search->length > 0) {
 761                                 $data["pubkey"] = $search->item(0)->nodeValue;
 762                                 if (strstr($data["pubkey"], 'RSA '))
 763                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 764                         }
 765
 766                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
 767                         if ($search->length > 0)
 768                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
 769                 }
 770
 771                 $avatar = array();
 772                 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
 773                 foreach ($photos AS $photo) {
 774                         $attr = array();
 775                         foreach ($photo->attributes as $attribute) {
 776                                 $attr[$attribute->name] = trim($attribute->value);
 777                         }
 778
 779                         if (isset($attr["src"]) AND isset($attr["width"])) {
 780                                 $avatar[$attr["width"]] = $attr["src"];
 781                         }
 782
 783                         // We don't have a width. So we just take everything that we got.
 784                         // This is a Hubzilla workaround which doesn't send a width.
 785                         if ((sizeof($avatar) == 0) AND isset($attr["src"])) {
 786                                 $avatar[] = $attr["src"];
 787                         }
 788                 }
 789
 790                 if (sizeof($avatar)) {
 791                         ksort($avatar);
 792                         $data["photo"] = array_pop($avatar);
 793                 }
 794
 795                 if ($dfrn) {
 796                         // Poll DFRN specific data
 797                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
 798                         if ($search->length > 0) {
 799                                 foreach ($search AS $link) {
 800                                         //$data["request"] = $search->item(0)->nodeValue;
 801                                         $attr = array();
 802                                         foreach ($link->attributes as $attribute)
 803                                                 $attr[$attribute->name] = trim($attribute->value);
 804
 805                                         $data[substr($attr["rel"], 5)] = $attr["href"];
 806                                 }
 807                         }
 808
 809                         // Older Friendica versions had used the "uid" field differently than newer versions
 810                         if ($data["nick"] == $data["guid"])
 811                                 unset($data["guid"]);
 812                 }
 813
 814
 815                 return $data;
 816         }
 817
 818         /**
 819          * @brief Check for Diaspora contact
 820          *
 821          * @param array $webfinger Webfinger data
 822          *
 823          * @return array Diaspora data
 824          */
 825         private function diaspora($webfinger) {
 826
 827                 $hcard = "";
 828                 $data = array();
 829                 foreach ($webfinger["links"] AS $link) {
 830                         if (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 831                                 $hcard = $link["href"];
 832                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 833                                 $data["baseurl"] = trim($link["href"], '/');
 834                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 835                                 $data["guid"] = $link["href"];
 836                         elseif (($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"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 840                                 $data["poll"] = $link["href"];
 841                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 842                                 $data["poco"] = $link["href"];
 843                         elseif (($link["rel"] == "salmon") AND ($link["href"] != ""))
 844                                 $data["notify"] = $link["href"];
 845                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 846                                 $data["pubkey"] = base64_decode($link["href"]);
 847
 848                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 849                                 if (strstr($data["pubkey"], 'RSA '))
 850                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 851                         }
 852                 }
 853
 854                 if (!isset($data["url"]) OR ($hcard == ""))
 855                         return false;
 856
 857                 if (is_array($webfinger["aliases"]))
 858                         foreach ($webfinger["aliases"] AS $alias)
 859                                 if (normalise_link($alias) != normalise_link($data["url"]) AND !strstr($alias, "@"))
 860                                         $data["alias"] = $alias;
 861
 862                 // Fetch further information from the hcard
 863                 $data = self::poll_hcard($hcard, $data);
 864
 865                 if (!$data)
 866                         return false;
 867
 868                 if (isset($data["url"]) AND isset($data["guid"]) AND isset($data["baseurl"]) AND
 869                         isset($data["pubkey"]) AND ($hcard != "")) {
 870                         $data["network"] = NETWORK_DIASPORA;
 871
 872                         // The Diaspora handle must always be lowercase
 873                         $data["addr"] = strtolower($data["addr"]);
 874
 875                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
 876                         $data["notify"] = $data["baseurl"]."/receive/users/".$data["guid"];
 877                         $data["batch"] = $data["baseurl"]."/receive/public";
 878                 } else
 879                         return false;
 880
 881                 return $data;
 882         }
 883
 884         /**
 885          * @brief Check for OStatus contact
 886          *
 887          * @param array $webfinger Webfinger data
 888          *
 889          * @return array OStatus data
 890          */
 891         private function ostatus($webfinger) {
 892                 $data = array();
 893                 if (is_array($webfinger["aliases"])) {
 894                         foreach ($webfinger["aliases"] AS $alias) {
 895                                 if (strstr($alias, "@")) {
 896                                         $data["addr"] = str_replace('acct:', '', $alias);
 897                                 }
 898                         }
 899                 }
 900
 901                 if (is_string($webfinger["subject"]) AND strstr($webfinger["subject"], "@")) {
 902                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
 903                 }
 904                 $pubkey = "";
 905                 foreach ($webfinger["links"] AS $link) {
 906                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 907                                 ($link["type"] == "text/html") AND ($link["href"] != "")) {
 908                                 $data["url"] = $link["href"];
 909                         } elseif (($link["rel"] == "salmon") AND ($link["href"] != "")) {
 910                                 $data["notify"] = $link["href"];
 911                         } elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != "")) {
 912                                 $data["poll"] = $link["href"];
 913                         } elseif (($link["rel"] == "magic-public-key") AND ($link["href"] != "")) {
 914                                 $pubkey = $link["href"];
 915
 916                                 if (substr($pubkey, 0, 5) === 'data:') {
 917                                         if (strstr($pubkey, ',')) {
 918                                                 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
 919                                         } else {
 920                                                 $pubkey = substr($pubkey, 5);
 921                                         }
 922                                 } elseif (normalise_link($pubkey) == 'http://') {
 923                                         $ret = z_fetch_url($pubkey);
 924                                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 925                                                 return false;
 926                                         }
 927                                         $pubkey = $ret['body'];
 928                                 }
 929
 930                                 $key = explode(".", $pubkey);
 931
 932                                 if (sizeof($key) >= 3) {
 933                                         $m = base64url_decode($key[1]);
 934                                         $e = base64url_decode($key[2]);
 935                                         $data["pubkey"] = metopem($m,$e);
 936                                 }
 937                         }
 938                 }
 939
 940                 if (isset($data["notify"]) AND isset($data["pubkey"]) AND
 941                         isset($data["poll"]) AND isset($data["url"])) {
 942                         $data["network"] = NETWORK_OSTATUS;
 943                 } else {
 944                         return false;
 945                 }
 946                 // Fetch all additional data from the feed
 947                 $ret = z_fetch_url($data["poll"]);
 948                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 949                         return false;
 950                 }
 951                 $feed = $ret['body'];
 952                 $feed_data = feed_import($feed,$dummy1,$dummy2, $dummy3, true);
 953                 if (!$feed_data) {
 954                         return false;
 955                 }
 956                 if ($feed_data["header"]["author-name"] != "") {
 957                         $data["name"] = $feed_data["header"]["author-name"];
 958                 }
 959                 if ($feed_data["header"]["author-nick"] != "") {
 960                         $data["nick"] = $feed_data["header"]["author-nick"];
 961                 }
 962                 if ($feed_data["header"]["author-avatar"] != "") {
 963                         $data["photo"] = ostatus::fix_avatar($feed_data["header"]["author-avatar"], $data["url"]);
 964                 }
 965                 if ($feed_data["header"]["author-id"] != "") {
 966                         $data["alias"] = $feed_data["header"]["author-id"];
 967                 }
 968                 if ($feed_data["header"]["author-location"] != "") {
 969                         $data["location"] = $feed_data["header"]["author-location"];
 970                 }
 971                 if ($feed_data["header"]["author-about"] != "") {
 972                         $data["about"] = $feed_data["header"]["author-about"];
 973                 }
 974                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
 975                 // So we take the value that we just fetched, although the other one worked as well
 976                 if ($feed_data["header"]["author-link"] != "") {
 977                         $data["url"] = $feed_data["header"]["author-link"];
 978                 }
 979                 /// @todo Fetch location and "about" from the feed as well
 980                 return $data;
 981         }
 982
 983         /**
 984          * @brief Fetch data from a pump.io profile page
 985          *
 986          * @param string $profile Link to the profile page
 987          *
 988          * @return array profile data
 989          */
 990         private function pumpio_profile_data($profile) {
 991
 992                 $doc = new DOMDocument();
 993                 if (!@$doc->loadHTMLFile($profile))
 994                         return false;
 995
 996                 $xpath = new DomXPath($doc);
 997
 998                 $data = array();
 999
1000                 // This is ugly - but pump.io doesn't seem to know a better way for it
1001                 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1002                 $pos = strpos($data["name"], chr(10));
1003                 if ($pos)
1004                         $data["name"] = trim(substr($data["name"], 0, $pos));
1005
1006                 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1007                 if ($avatar)
1008                         foreach ($avatar->attributes as $attribute)
1009                                 if ($attribute->name == "src")
1010                                         $data["photo"] = trim($attribute->value);
1011
1012                 $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
1013                 $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
1014
1015                 return $data;
1016         }
1017
1018         /**
1019          * @brief Check for pump.io contact
1020          *
1021          * @param array $webfinger Webfinger data
1022          *
1023          * @return array pump.io data
1024          */
1025         private function pumpio($webfinger) {
1026
1027                 $data = array();
1028                 foreach ($webfinger["links"] AS $link) {
1029                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
1030                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
1031                                 $data["url"] = $link["href"];
1032                         elseif (($link["rel"] == "activity-inbox") AND ($link["href"] != ""))
1033                                 $data["notify"] = $link["href"];
1034                         elseif (($link["rel"] == "activity-outbox") AND ($link["href"] != ""))
1035                                 $data["poll"] = $link["href"];
1036                         elseif (($link["rel"] == "dialback") AND ($link["href"] != ""))
1037                                 $data["dialback"] = $link["href"];
1038                 }
1039                 if (isset($data["poll"]) AND isset($data["notify"]) AND
1040                         isset($data["dialback"]) AND isset($data["url"])) {
1041
1042                         // by now we use these fields only for the network type detection
1043                         // So we unset all data that isn't used at the moment
1044                         unset($data["dialback"]);
1045
1046                         $data["network"] = NETWORK_PUMPIO;
1047                 } else
1048                         return false;
1049
1050                 $profile_data = self::pumpio_profile_data($data["url"]);
1051
1052                 if (!$profile_data)
1053                         return false;
1054
1055                 $data = array_merge($data, $profile_data);
1056
1057                 return $data;
1058         }
1059
1060         /**
1061          * @brief Check page for feed link
1062          *
1063          * @param string $url Page link
1064          *
1065          * @return string feed link
1066          */
1067         private function get_feed_link($url) {
1068                 $doc = new DOMDocument();
1069
1070                 if (!@$doc->loadHTMLFile($url))
1071                         return false;
1072
1073                 $xpath = new DomXPath($doc);
1074
1075                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1076                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1077                 if (!is_object($feeds))
1078                         return false;
1079
1080                 if ($feeds->length == 0)
1081                         return false;
1082
1083                 $feed_url = "";
1084
1085                 foreach ($feeds AS $feed) {
1086                         $attr = array();
1087                         foreach ($feed->attributes as $attribute)
1088                         $attr[$attribute->name] = trim($attribute->value);
1089
1090                         if ($feed_url == "")
1091                                 $feed_url = $attr["href"];
1092                 }
1093
1094                 return $feed_url;
1095         }
1096
1097         /**
1098          * @brief Check for feed contact
1099          *
1100          * @param string $url Profile link
1101          * @param boolean $probe Do a probe if the page contains a feed link
1102          *
1103          * @return array feed data
1104          */
1105         private function feed($url, $probe = true) {
1106                 $ret = z_fetch_url($url);
1107                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1108                         return false;
1109                 }
1110                 $feed = $ret['body'];
1111                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1112
1113                 if (!$feed_data) {
1114                         if (!$probe)
1115                                 return false;
1116
1117                         $feed_url = self::get_feed_link($url);
1118
1119                         if (!$feed_url)
1120                                 return false;
1121
1122                         return self::feed($feed_url, false);
1123                 }
1124
1125                 if ($feed_data["header"]["author-name"] != "")
1126                         $data["name"] = $feed_data["header"]["author-name"];
1127
1128                 if ($feed_data["header"]["author-nick"] != "")
1129                         $data["nick"] = $feed_data["header"]["author-nick"];
1130
1131                 if ($feed_data["header"]["author-avatar"] != "")
1132                         $data["photo"] = $feed_data["header"]["author-avatar"];
1133
1134                 if ($feed_data["header"]["author-id"] != "")
1135                         $data["alias"] = $feed_data["header"]["author-id"];
1136
1137                 $data["url"] = $url;
1138                 $data["poll"] = $url;
1139
1140                 if ($feed_data["header"]["author-link"] != "")
1141                         $data["baseurl"] = $feed_data["header"]["author-link"];
1142                 else
1143                         $data["baseurl"] = $data["url"];
1144
1145                 $data["network"] = NETWORK_FEED;
1146
1147                 return $data;
1148         }
1149
1150         /**
1151          * @brief Check for mail contact
1152          *
1153          * @param string $uri Profile link
1154          * @param integer $uid User ID
1155          *
1156          * @return array mail data
1157          */
1158         private function mail($uri, $uid) {
1159
1160                 if (!validate_email($uri))
1161                         return false;
1162
1163                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1164
1165                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1166
1167                 if (dbm::is_result($x) && dbm::is_result($r)) {
1168                         $mailbox = construct_mailbox_name($r[0]);
1169                         $password = '';
1170                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password,$x[0]['prvkey']);
1171                         $mbox = email_connect($mailbox,$r[0]['user'], $password);
1172                         if(!mbox)
1173                                 return false;
1174                 }
1175
1176                 $msgs = email_poll($mbox, $uri);
1177                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1178
1179                 if (!count($msgs))
1180                         return false;
1181
1182                 $data = array();
1183
1184                 $data["addr"] = $uri;
1185                 $data["network"] = NETWORK_MAIL;
1186                 $data["name"] = substr($uri, 0, strpos($uri,'@'));
1187                 $data["nick"] = $data["name"];
1188                 $data["photo"] = avatar_img($uri);
1189
1190                 $phost = substr($uri, strpos($uri,'@') + 1);
1191                 $data["url"] = 'http://'.$phost."/".$data["nick"];
1192                 $data["notify"] = 'smtp '.random_string();
1193                 $data["poll"] = 'email '.random_string();
1194
1195                 $x = email_msg_meta($mbox, $msgs[0]);
1196                 if(stristr($x[0]->from, $uri))
1197                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1198                 elseif(stristr($x[0]->to, $uri))
1199                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1200                 if(isset($adr)) {
1201                         foreach($adr as $feadr) {
1202                                 if((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1203                                         &&(strcasecmp($feadr->host, $phost) == 0)
1204                                         && (strlen($feadr->personal))) {
1205
1206                                         $personal = imap_mime_header_decode($feadr->personal);
1207                                         $data["name"] = "";
1208                                         foreach($personal as $perspart)
1209                                                 if ($perspart->charset != "default")
1210                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1211                                                 else
1212                                                         $data["name"] .= $perspart->text;
1213
1214                                         $data["name"] = notags($data["name"]);
1215                                 }
1216                         }
1217                 }
1218                 imap_close($mbox);
1219
1220                 return $data;
1221         }
1222 }