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