include/Probe.php

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