include/Probe.php

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