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