src/Network/Probe.php

   1 <?php
   2
   3 namespace Friendica\Network;
   4
   5 /**
   6  * @file src/Network/Probe.php
   7  * @brief Functions for probing URL
   8  *
   9  */
  10
  11 use Friendica\App;
  12 use Friendica\Core\Config;
  13
  14 use dbm;
  15 use Cache;
  16 use xml;
  17
  18 use DomXPath;
  19 use DOMDocument;
  20
  21 require_once 'include/feed.php';
  22 require_once 'include/email.php';
  23 require_once 'include/network.php';
  24
  25 /**
  26  * @brief This class contain functions for probing URL
  27  *
  28  */
  29 class Probe {
  30
  31         private static $baseurl;
  32
  33         /**
  34          * @brief Rearrange the array so that it always has the same order
  35          *
  36          * @param array $data Unordered data
  37          *
  38          * @return array Ordered data
  39          */
  40         private static function rearrangeData($data) {
  41                 $fields = array("name", "nick", "guid", "url", "addr", "alias",
  42                                 "photo", "community", "keywords", "location", "about",
  43                                 "batch", "notify", "poll", "request", "confirm", "poco",
  44                                 "priority", "network", "pubkey", "baseurl");
  45
  46                 $newdata = array();
  47                 foreach ($fields as $field) {
  48                         if (isset($data[$field])) {
  49                                 $newdata[$field] = $data[$field];
  50                         } else {
  51                                 $newdata[$field] = "";
  52                         }
  53                 }
  54
  55                 // We don't use the "priority" field anymore and replace it with a dummy.
  56                 $newdata["priority"] = 0;
  57
  58                 return $newdata;
  59         }
  60
  61         /**
  62          * @brief Check if the hostname belongs to the own server
  63          *
  64          * @param string $host The hostname that is to be checked
  65          *
  66          * @return bool Does the testes hostname belongs to the own server?
  67          */
  68         private static function ownHost($host) {
  69                 $own_host = get_app()->get_hostname();
  70
  71                 $parts = parse_url($host);
  72
  73                 if (!isset($parts['scheme'])) {
  74                         $parts = parse_url('http://'.$host);
  75                 }
  76
  77                 if (!isset($parts['host'])) {
  78                         return false;
  79                 }
  80                 return $parts['host'] == $own_host;
  81         }
  82
  83         /**
  84          * @brief Probes for XRD data
  85          *
  86          * @param string $host The host part of an url
  87          *
  88          * @return array
  89          *      'lrdd' => Link to LRDD endpoint
  90          *      'lrdd-xml' => Link to LRDD endpoint in XML format
  91          *      'lrdd-json' => Link to LRDD endpoint in JSON format
  92          */
  93         private static function xrd($host) {
  94
  95                 // Reset the static variable
  96                 self::$baseurl = '';
  97
  98                 $ssl_url = "https://".$host."/.well-known/host-meta";
  99                 $url = "http://".$host."/.well-known/host-meta";
 100
 101                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 102                 $redirects = 0;
 103
 104                 logger("Probing for ".$host, LOGGER_DEBUG);
 105
 106                 $ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 107                 if ($ret['success']) {
 108                         $xml = $ret['body'];
 109                         $xrd = parse_xml_string($xml, false);
 110                 }
 111
 112                 if (!is_object($xrd)) {
 113                         $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 114                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 115                                 logger("Probing timeout for ".$url, LOGGER_DEBUG);
 116                                 return false;
 117                         }
 118                         $xml = $ret['body'];
 119                         $xrd = parse_xml_string($xml, false);
 120                 }
 121                 if (!is_object($xrd)) {
 122                         logger("No xrd object found for ".$host, LOGGER_DEBUG);
 123                         return array();
 124                 }
 125
 126                 $links = xml::element_to_array($xrd);
 127                 if (!isset($links["xrd"]["link"])) {
 128                         logger("No xrd data found for ".$host, LOGGER_DEBUG);
 129                         return array();
 130                 }
 131
 132                 $xrd_data = array();
 133
 134                 foreach ($links["xrd"]["link"] as $value => $link) {
 135                         if (!empty($link["@attributes"])) {
 136                                 $attributes = $link["@attributes"];
 137                         } elseif ($value == "@attributes") {
 138                                 $attributes = $link;
 139                         } else {
 140                                 continue;
 141                         }
 142
 143                         if (($attributes["rel"] == "lrdd")
 144                                 && ($attributes["type"] == "application/xrd+xml")
 145                         ) {
 146                                 $xrd_data["lrdd-xml"] = $attributes["template"];
 147                         } elseif (($attributes["rel"] == "lrdd")
 148                                 && ($attributes["type"] == "application/json")
 149                         ) {
 150                                 $xrd_data["lrdd-json"] = $attributes["template"];
 151                         } elseif ($attributes["rel"] == "lrdd") {
 152                                 $xrd_data["lrdd"] = $attributes["template"];
 153                         }
 154                 }
 155
 156                 self::$baseurl = "http://".$host;
 157
 158                 logger("Probing successful for ".$host, LOGGER_DEBUG);
 159
 160                 return $xrd_data;
 161         }
 162
 163         /**
 164          * @brief Perform Webfinger lookup and return DFRN data
 165          *
 166          * Given an email style address, perform webfinger lookup and
 167          * return the resulting DFRN profile URL, or if no DFRN profile URL
 168          * is located, returns an OStatus subscription template (prefixed
 169          * with the string 'stat:' to identify it as on OStatus template).
 170          * If this isn't an email style address just return $webbie.
 171          * Return an empty string if email-style addresses but webfinger fails,
 172          * or if the resultant personal XRD doesn't contain a supported
 173          * subscription/friend-request attribute.
 174          *
 175          * amended 7/9/2011 to return an hcard which could save potentially loading
 176          * a lengthy content page to scrape dfrn attributes
 177          *
 178          * @param string $webbie Address that should be probed
 179          * @param string $hcard_url Link to the hcard - is returned by reference
 180          *
 181          * @return string profile link
 182          */
 183         public static function webfingerDfrn($webbie, &$hcard_url) {
 184
 185                 $profile_link = '';
 186
 187                 $links = self::lrdd($webbie);
 188                 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
 189                 if (count($links)) {
 190                         foreach ($links as $link) {
 191                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
 192                                         $profile_link = $link['@attributes']['href'];
 193                                 }
 194                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
 195                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 196                                 }
 197                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 198                                         $hcard_url = $link['@attributes']['href'];
 199                                 }
 200                         }
 201                 }
 202                 return $profile_link;
 203         }
 204
 205         /**
 206          * @brief Check an URI for LRDD data
 207          *
 208          * this is a replacement for the "lrdd" function in include/network.php.
 209          * It isn't used in this class and has some redundancies in the code.
 210          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 211          *
 212          * @param string $uri Address that should be probed
 213          *
 214          * @return array uri data
 215          */
 216         public static function lrdd($uri) {
 217
 218                 $lrdd = self::xrd($uri);
 219                 $webfinger = null;
 220
 221                 if (is_bool($lrdd)) {
 222                         return array();
 223                 }
 224
 225                 if (!$lrdd) {
 226                         $parts = @parse_url($uri);
 227                         if (!$parts) {
 228                                 return array();
 229                         }
 230
 231                         $host = $parts["host"];
 232                         if (!empty($parts["port"])) {
 233                                 $host .= ':'.$parts["port"];
 234                         }
 235
 236                         $path_parts = explode("/", trim($parts["path"], "/"));
 237
 238                         $nick = array_pop($path_parts);
 239
 240                         do {
 241                                 $lrdd = self::xrd($host);
 242                                 $host .= "/".array_shift($path_parts);
 243                         } while (!$lrdd && (sizeof($path_parts) > 0));
 244                 }
 245
 246                 if (!$lrdd) {
 247                         logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
 248                         return array();
 249                 }
 250
 251                 foreach ($lrdd as $key => $link) {
 252                         if ($webfinger) {
 253                                 continue;
 254                         }
 255
 256                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
 257                                 continue;
 258                         }
 259
 260                         $path = str_replace('{uri}', urlencode($uri), $link);
 261                         $webfinger = self::webfinger($path);
 262
 263                         if (!$webfinger && (strstr($uri, "@"))) {
 264                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
 265                                 $webfinger = self::webfinger($path);
 266                         }
 267
 268                         // Special treatment for Mastodon
 269                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
 270                         // But the webfinger for this format fails.
 271                         if (!$webfinger && !empty($nick)) {
 272                                 // Mastodon uses a "@" as prefix for usernames in their url format
 273                                 $nick = ltrim($nick, '@');
 274
 275                                 $addr = $nick."@".$host;
 276
 277                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 278                                 $webfinger = self::webfinger($path);
 279                         }
 280                 }
 281
 282                 if (!is_array($webfinger["links"])) {
 283                         logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
 284                         return false;
 285                 }
 286
 287                 $data = array();
 288
 289                 foreach ($webfinger["links"] as $link) {
 290                         $data[] = array("@attributes" => $link);
 291                 }
 292
 293                 if (is_array($webfinger["aliases"])) {
 294                         foreach ($webfinger["aliases"] as $alias) {
 295                                 $data[] = array("@attributes" =>
 296                                                         array("rel" => "alias",
 297                                                                 "href" => $alias));
 298                         }
 299                 }
 300
 301                 return $data;
 302         }
 303
 304         /**
 305          * @brief Fetch information (protocol endpoints and user information) about a given uri
 306          *
 307          * @param string $uri Address that should be probed
 308          * @param string $network Test for this specific network
 309          * @param integer $uid User ID for the probe (only used for mails)
 310          * @param boolean $cache Use cached values?
 311          *
 312          * @return array uri data
 313          */
 314         public static function uri($uri, $network = "", $uid = 0, $cache = true) {
 315
 316                 if ($cache) {
 317                         $result = Cache::get("probe_url:".$network.":".$uri);
 318                         if (!is_null($result)) {
 319                                 return $result;
 320                         }
 321                 }
 322
 323                 if ($uid == 0) {
 324                         $uid = local_user();
 325                 }
 326
 327                 $data = self::detect($uri, $network, $uid);
 328
 329                 if (!isset($data["url"])) {
 330                         $data["url"] = $uri;
 331                 }
 332
 333                 if ($data["photo"] != "") {
 334                         $data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
 335                 } else {
 336                         $data["photo"] = App::get_baseurl().'/images/person-175.jpg';
 337                 }
 338
 339                 if (empty($data["name"])) {
 340                         if (!empty($data["nick"])) {
 341                                 $data["name"] = $data["nick"];
 342                         }
 343
 344                         if ($data["name"] == "") {
 345                                 $data["name"] = $data["url"];
 346                         }
 347                 }
 348
 349                 if (empty($data["nick"])) {
 350                         $data["nick"] = strtolower($data["name"]);
 351
 352                         if (strpos($data['nick'], ' ')) {
 353                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 354                         }
 355                 }
 356
 357                 if (self::$baseurl != "") {
 358                         $data["baseurl"] = self::$baseurl;
 359                 }
 360
 361                 if (!isset($data["network"])) {
 362                         $data["network"] = NETWORK_PHANTOM;
 363                 }
 364
 365                 $data = self::rearrangeData($data);
 366
 367                 // Only store into the cache if the value seems to be valid
 368                 if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
 369                         Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
 370
 371                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 372                         /// The biggest problem is the avatar picture that could have a reduced image size.
 373                         /// It should only be updated if the existing picture isn't existing anymore.
 374                         /// We only update the contact when it is no probing for a specific network.
 375                         if (($data['network'] != NETWORK_FEED)
 376                                 && ($network == "")
 377                                 && $data["name"]
 378                                 && $data["nick"]
 379                                 && $data["url"]
 380                                 && $data["addr"]
 381                                 && $data["poll"]
 382                         ) {
 383                                 q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 384                                                 `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 385                                         WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 386                                         dbesc($data["name"]),
 387                                         dbesc($data["nick"]),
 388                                         dbesc($data["url"]),
 389                                         dbesc($data["addr"]),
 390                                         dbesc($data["notify"]),
 391                                         dbesc($data["poll"]),
 392                                         dbesc($data["alias"]),
 393                                         dbesc(datetime_convert()),
 394                                         dbesc(normalise_link($data['url']))
 395                                 );
 396                         }
 397                 }
 398
 399                 return $data;
 400         }
 401
 402         /**
 403          * @brief Switch the scheme of an url between http and https
 404          *
 405          * @param string $url URL
 406          *
 407          * @return string switched URL
 408          */
 409         private static function switchScheme($url) {
 410                 $parts = parse_url($url);
 411
 412                 if (!isset($parts['scheme'])) {
 413                         return $url;
 414                 }
 415
 416                 if ($parts['scheme'] == 'http') {
 417                         $url = str_replace('http://', 'https://', $url);
 418                 } elseif ($parts['scheme'] == 'https') {
 419                         $url = str_replace('https://', 'http://', $url);
 420                 }
 421
 422                 return $url;
 423         }
 424
 425         /**
 426          * @brief Checks if a profile url should be OStatus but only provides partial information
 427          *
 428          * @param array $webfinger Webfinger data
 429          * @param string $lrdd Path template for webfinger request
 430          *
 431          * @return array fixed webfinger data
 432          */
 433         private static function fixOstatus($webfinger, $lrdd) {
 434                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
 435                         return $webfinger;
 436                 }
 437
 438                 $is_ostatus = false;
 439                 $has_key = false;
 440
 441                 foreach ($webfinger['links'] as $link) {
 442                         if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
 443                                 $is_ostatus = true;
 444                         }
 445                         if ($link['rel'] == 'magic-public-key') {
 446                                 $has_key = true;
 447                         }
 448                 }
 449
 450                 if (!$is_ostatus || $has_key) {
 451                         return $webfinger;
 452                 }
 453
 454                 $url = self::switchScheme($webfinger['subject']);
 455                 $path = str_replace('{uri}', urlencode($url), $lrdd);
 456                 $webfinger2 = self::webfinger($path);
 457
 458                 // Is the new webfinger detectable as OStatus?
 459                 if (self::ostatus($webfinger2, true)) {
 460                         $webfinger = $webfinger2;
 461                 }
 462
 463                 return $webfinger;
 464         }
 465
 466         /**
 467          * @brief Fetch information (protocol endpoints and user information) about a given uri
 468          *
 469          * This function is only called by the "uri" function that adds caching and rearranging of data.
 470          *
 471          * @param string $uri Address that should be probed
 472          * @param string $network Test for this specific network
 473          * @param integer $uid User ID for the probe (only used for mails)
 474          *
 475          * @return array uri data
 476          */
 477         private static function detect($uri, $network, $uid) {
 478                 $parts = parse_url($uri);
 479
 480                 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
 481                         $host = $parts["host"];
 482                         if (!empty($parts["port"])) {
 483                                 $host .= ':'.$parts["port"];
 484                         }
 485
 486                         if ($host == 'twitter.com') {
 487                                 return array("network" => NETWORK_TWITTER);
 488                         }
 489                         $lrdd = self::xrd($host);
 490
 491                         if (is_bool($lrdd)) {
 492                                 return array();
 493                         }
 494
 495                         $path_parts = explode("/", trim($parts["path"], "/"));
 496
 497                         while (!$lrdd && (sizeof($path_parts) > 1)) {
 498                                 $host .= "/".array_shift($path_parts);
 499                                 $lrdd = self::xrd($host);
 500                         }
 501                         if (!$lrdd) {
 502                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 503                                 return self::feed($uri);
 504                         }
 505                         $nick = array_pop($path_parts);
 506
 507                         // Mastodon uses a "@" as prefix for usernames in their url format
 508                         $nick = ltrim($nick, '@');
 509
 510                         $addr = $nick."@".$host;
 511
 512                 } elseif (strstr($uri, '@')) {
 513                         // If the URI starts with "mailto:" then jump directly to the mail detection
 514                         if (strpos($uri, 'mailto:') !== false) {
 515                                 $uri = str_replace('mailto:', '', $uri);
 516                                 return self::mail($uri, $uid);
 517                         }
 518
 519                         if ($network == NETWORK_MAIL) {
 520                                 return self::mail($uri, $uid);
 521                         }
 522                         // Remove "acct:" from the URI
 523                         $uri = str_replace('acct:', '', $uri);
 524
 525                         $host = substr($uri, strpos($uri, '@') + 1);
 526                         $nick = substr($uri, 0, strpos($uri, '@'));
 527
 528                         if (strpos($uri, '@twitter.com')) {
 529                                 return array("network" => NETWORK_TWITTER);
 530                         }
 531                         $lrdd = self::xrd($host);
 532
 533                         if (is_bool($lrdd)) {
 534                                 return array();
 535                         }
 536
 537                         if (!$lrdd) {
 538                                 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
 539                                 return self::mail($uri, $uid);
 540                         }
 541                         $addr = $uri;
 542
 543                 } else {
 544                         logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
 545                         return false;
 546                 }
 547
 548                 $webfinger = false;
 549
 550                 /// @todo Do we need the prefix "acct:" or "acct://"?
 551
 552                 foreach ($lrdd as $key => $link) {
 553                         if ($webfinger) {
 554                                 continue;
 555                         }
 556                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
 557                                 continue;
 558                         }
 559                         // At first try it with the given uri
 560                         $path = str_replace('{uri}', urlencode($uri), $link);
 561                         $webfinger = self::webfinger($path);
 562
 563                         // Fix possible problems with GNU Social probing to wrong scheme
 564                         $webfinger = self::fixOstatus($webfinger, $link);
 565
 566                         // We cannot be sure that the detected address was correct, so we don't use the values
 567                         if ($webfinger && ($uri != $addr)) {
 568                                 $nick = "";
 569                                 $addr = "";
 570                         }
 571
 572                         // Try webfinger with the address (user@domain.tld)
 573                         if (!$webfinger) {
 574                                 $path = str_replace('{uri}', urlencode($addr), $link);
 575                                 $webfinger = self::webfinger($path);
 576                         }
 577
 578                         // Mastodon needs to have it with "acct:"
 579                         if (!$webfinger) {
 580                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 581                                 $webfinger = self::webfinger($path);
 582                         }
 583                 }
 584                 if (!$webfinger) {
 585                         return self::feed($uri);
 586                 }
 587
 588                 $result = false;
 589
 590                 logger("Probing ".$uri, LOGGER_DEBUG);
 591
 592                 if (in_array($network, array("", NETWORK_DFRN))) {
 593                         $result = self::dfrn($webfinger);
 594                 }
 595                 if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
 596                         $result = self::diaspora($webfinger);
 597                 }
 598                 if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
 599                         $result = self::ostatus($webfinger);
 600                 }
 601                 if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
 602                         $result = self::pumpio($webfinger);
 603                 }
 604                 if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
 605                         $result = self::feed($uri);
 606                 } else {
 607                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 608                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 609                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 610                                 $result["nick"] = $nick;
 611                         }
 612
 613                         if (empty($result["addr"]) && ($addr != "")) {
 614                                 $result["addr"] = $addr;
 615                         }
 616                 }
 617
 618                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 619
 620                 if (empty($result["baseurl"])) {
 621                         $pos = strpos($result["url"], $host);
 622                         if ($pos) {
 623                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 624                         }
 625                 }
 626
 627                 return $result;
 628         }
 629
 630         /**
 631          * @brief Perform a webfinger request.
 632          *
 633          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 634          *
 635          * @param string $url Address that should be probed
 636          *
 637          * @return array webfinger data
 638          */
 639         private static function webfinger($url) {
 640
 641                 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
 642                 $redirects = 0;
 643
 644                 $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 645                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 646                         return false;
 647                 }
 648                 $data = $ret['body'];
 649
 650                 $xrd = parse_xml_string($data, false);
 651
 652                 if (!is_object($xrd)) {
 653                         // If it is not XML, maybe it is JSON
 654                         $webfinger = json_decode($data, true);
 655
 656                         if (!isset($webfinger["links"])) {
 657                                 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
 658                                 return false;
 659                         }
 660
 661                         return $webfinger;
 662                 }
 663
 664                 $xrd_arr = xml::element_to_array($xrd);
 665                 if (!isset($xrd_arr["xrd"]["link"])) {
 666                         logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
 667                         return false;
 668                 }
 669
 670                 $webfinger = array();
 671
 672                 if (!empty($xrd_arr["xrd"]["subject"])) {
 673                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 674                 }
 675
 676                 if (!empty($xrd_arr["xrd"]["alias"])) {
 677                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 678                 }
 679
 680                 $webfinger["links"] = array();
 681
 682                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 683                         if (!empty($data["@attributes"])) {
 684                                 $attributes = $data["@attributes"];
 685                         } elseif ($value == "@attributes") {
 686                                 $attributes = $data;
 687                         } else {
 688                                 continue;
 689                         }
 690
 691                         $webfinger["links"][] = $attributes;
 692                 }
 693                 return $webfinger;
 694         }
 695
 696         /**
 697          * @brief Poll the Friendica specific noscrape page.
 698          *
 699          * "noscrape" is a faster alternative to fetch the data from the hcard.
 700          * This functionality was originally created for the directory.
 701          *
 702          * @param string $noscrape_url Link to the noscrape page
 703          * @param array $data The already fetched data
 704          *
 705          * @return array noscrape data
 706          */
 707         private static function pollNoscrape($noscrape_url, $data) {
 708                 $ret = z_fetch_url($noscrape_url);
 709                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 710                         return false;
 711                 }
 712                 $content = $ret['body'];
 713                 if (!$content) {
 714                         logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
 715                         return false;
 716                 }
 717
 718                 $json = json_decode($content, true);
 719                 if (!is_array($json)) {
 720                         logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
 721                         return false;
 722                 }
 723
 724                 if (!empty($json["fn"])) {
 725                         $data["name"] = $json["fn"];
 726                 }
 727
 728                 if (!empty($json["addr"])) {
 729                         $data["addr"] = $json["addr"];
 730                 }
 731
 732                 if (!empty($json["nick"])) {
 733                         $data["nick"] = $json["nick"];
 734                 }
 735
 736                 if (!empty($json["comm"])) {
 737                         $data["community"] = $json["comm"];
 738                 }
 739
 740                 if (!empty($json["tags"])) {
 741                         $keywords = implode(" ", $json["tags"]);
 742                         if ($keywords != "") {
 743                                 $data["keywords"] = $keywords;
 744                         }
 745                 }
 746
 747                 $location = formatted_location($json);
 748                 if ($location) {
 749                         $data["location"] = $location;
 750                 }
 751
 752                 if (!empty($json["about"])) {
 753                         $data["about"] = $json["about"];
 754                 }
 755
 756                 if (!empty($json["key"])) {
 757                         $data["pubkey"] = $json["key"];
 758                 }
 759
 760                 if (!empty($json["photo"])) {
 761                         $data["photo"] = $json["photo"];
 762                 }
 763
 764                 if (!empty($json["dfrn-request"])) {
 765                         $data["request"] = $json["dfrn-request"];
 766                 }
 767
 768                 if (!empty($json["dfrn-confirm"])) {
 769                         $data["confirm"] = $json["dfrn-confirm"];
 770                 }
 771
 772                 if (!empty($json["dfrn-notify"])) {
 773                         $data["notify"] = $json["dfrn-notify"];
 774                 }
 775
 776                 if (!empty($json["dfrn-poll"])) {
 777                         $data["poll"] = $json["dfrn-poll"];
 778                 }
 779
 780                 return $data;
 781         }
 782
 783         /**
 784          * @brief Check for valid DFRN data
 785          *
 786          * @param array $data DFRN data
 787          *
 788          * @return int Number of errors
 789          */
 790         public static function validDfrn($data) {
 791                 $errors = 0;
 792                 if (!isset($data['key'])) {
 793                         $errors ++;
 794                 }
 795                 if (!isset($data['dfrn-request'])) {
 796                         $errors ++;
 797                 }
 798                 if (!isset($data['dfrn-confirm'])) {
 799                         $errors ++;
 800                 }
 801                 if (!isset($data['dfrn-notify'])) {
 802                         $errors ++;
 803                 }
 804                 if (!isset($data['dfrn-poll'])) {
 805                         $errors ++;
 806                 }
 807                 return $errors;
 808         }
 809
 810         /**
 811          * @brief Fetch data from a DFRN profile page and via "noscrape"
 812          *
 813          * @param string $profile_link Link to the profile page
 814          *
 815          * @return array profile data
 816          */
 817         public static function profile($profile_link) {
 818
 819                 $data = array();
 820
 821                 logger("Check profile ".$profile_link, LOGGER_DEBUG);
 822
 823                 // Fetch data via noscrape - this is faster
 824                 $noscrape_url = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile_link);
 825                 $data = self::pollNoscrape($noscrape_url, $data);
 826
 827                 if (!isset($data["notify"])
 828                         || !isset($data["confirm"])
 829                         || !isset($data["request"])
 830                         || !isset($data["poll"])
 831                         || !isset($data["poco"])
 832                         || !isset($data["name"])
 833                         || !isset($data["photo"])
 834                 ) {
 835                         $data = self::pollHcard($profile_link, $data, true);
 836                 }
 837
 838                 $prof_data = array();
 839                 $prof_data["addr"]         = $data["addr"];
 840                 $prof_data["nick"]         = $data["nick"];
 841                 $prof_data["dfrn-request"] = $data["request"];
 842                 $prof_data["dfrn-confirm"] = $data["confirm"];
 843                 $prof_data["dfrn-notify"]  = $data["notify"];
 844                 $prof_data["dfrn-poll"]    = $data["poll"];
 845                 $prof_data["dfrn-poco"]    = $data["poco"];
 846                 $prof_data["photo"]        = $data["photo"];
 847                 $prof_data["fn"]           = $data["name"];
 848                 $prof_data["key"]          = $data["pubkey"];
 849
 850                 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
 851
 852                 return $prof_data;
 853         }
 854
 855         /**
 856          * @brief Check for DFRN contact
 857          *
 858          * @param array $webfinger Webfinger data
 859          *
 860          * @return array DFRN data
 861          */
 862         private static function dfrn($webfinger) {
 863
 864                 $hcard_url = "";
 865                 $data = array();
 866                 foreach ($webfinger["links"] as $link) {
 867                         if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
 868                                 $data["network"] = NETWORK_DFRN;
 869                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
 870                                 $data["poll"] = $link["href"];
 871                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
 872                                 $data["url"] = $link["href"];
 873                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
 874                                 $hcard_url = $link["href"];
 875                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
 876                                 $data["poco"] = $link["href"];
 877                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
 878                                 $data["photo"] = $link["href"];
 879                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
 880                                 $data["baseurl"] = trim($link["href"], '/');
 881                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
 882                                 $data["guid"] = $link["href"];
 883                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
 884                                 $data["pubkey"] = base64_decode($link["href"]);
 885
 886                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
 887                                 if (strstr($data["pubkey"], 'RSA ')) {
 888                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 889                                 }
 890                         }
 891                 }
 892
 893                 if (is_array($webfinger["aliases"])) {
 894                         foreach ($webfinger["aliases"] as $alias) {
 895                                 if (substr($alias, 0, 5) == 'acct:') {
 896                                         $data["addr"] = substr($alias, 5);
 897                                 }
 898                         }
 899                 }
 900
 901                 if (!isset($data["network"]) || ($hcard_url == "")) {
 902                         return false;
 903                 }
 904
 905                 // Fetch data via noscrape - this is faster
 906                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
 907                 $data = self::pollNoscrape($noscrape_url, $data);
 908
 909                 if (isset($data["notify"])
 910                         && isset($data["confirm"])
 911                         && isset($data["request"])
 912                         && isset($data["poll"])
 913                         && isset($data["name"])
 914                         && isset($data["photo"])
 915                 ) {
 916                         return $data;
 917                 }
 918
 919                 $data = self::pollHcard($hcard_url, $data, true);
 920
 921                 return $data;
 922         }
 923
 924         /**
 925          * @brief Poll the hcard page (Diaspora and Friendica specific)
 926          *
 927          * @param string $hcard_url Link to the hcard page
 928          * @param array $data The already fetched data
 929          * @param boolean $dfrn Poll DFRN specific data
 930          *
 931          * @return array hcard data
 932          */
 933         private static function pollHcard($hcard_url, $data, $dfrn = false) {
 934                 $ret = z_fetch_url($hcard_url);
 935                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 936                         return false;
 937                 }
 938                 $content = $ret['body'];
 939                 if (!$content) {
 940                         return false;
 941                 }
 942
 943                 $doc = new DOMDocument();
 944                 if (!@$doc->loadHTML($content)) {
 945                         return false;
 946                 }
 947
 948                 $xpath = new DomXPath($doc);
 949
 950                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 951                 if (!is_object($vcards)) {
 952                         return false;
 953                 }
 954
 955                 if ($vcards->length > 0) {
 956                         $vcard = $vcards->item(0);
 957
 958                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 959                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 960                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 961                         if (($search->length > 0) && ($data["guid"] == "")) {
 962                                 $data["guid"] = $search->item(0)->nodeValue;
 963                         }
 964
 965                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 966                         if ($search->length > 0) {
 967                                 $data["nick"] = $search->item(0)->nodeValue;
 968                         }
 969
 970                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 971                         if ($search->length > 0) {
 972                                 $data["name"] = $search->item(0)->nodeValue;
 973                         }
 974
 975                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
 976                         if ($search->length > 0) {
 977                                 $data["searchable"] = $search->item(0)->nodeValue;
 978                         }
 979
 980                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
 981                         if ($search->length > 0) {
 982                                 $data["pubkey"] = $search->item(0)->nodeValue;
 983                                 if (strstr($data["pubkey"], 'RSA ')) {
 984                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 985                                 }
 986                         }
 987
 988                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
 989                         if ($search->length > 0) {
 990                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
 991                         }
 992                 }
 993
 994                 $avatar = array();
 995                 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
 996                 foreach ($photos as $photo) {
 997                         $attr = array();
 998                         foreach ($photo->attributes as $attribute) {
 999                                 $attr[$attribute->name] = trim($attribute->value);
1000                         }
1001
1002                         if (isset($attr["src"]) && isset($attr["width"])) {
1003                                 $avatar[$attr["width"]] = $attr["src"];
1004                         }
1005
1006                         // We don't have a width. So we just take everything that we got.
1007                         // This is a Hubzilla workaround which doesn't send a width.
1008                         if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1009                                 $avatar[] = $attr["src"];
1010                         }
1011                 }
1012
1013                 if (sizeof($avatar)) {
1014                         ksort($avatar);
1015                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1016                 }
1017
1018                 if ($dfrn) {
1019                         // Poll DFRN specific data
1020                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1021                         if ($search->length > 0) {
1022                                 foreach ($search as $link) {
1023                                         //$data["request"] = $search->item(0)->nodeValue;
1024                                         $attr = array();
1025                                         foreach ($link->attributes as $attribute) {
1026                                                 $attr[$attribute->name] = trim($attribute->value);
1027                                         }
1028
1029                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1030                                 }
1031                         }
1032
1033                         // Older Friendica versions had used the "uid" field differently than newer versions
1034                         if ($data["nick"] == $data["guid"]) {
1035                                 unset($data["guid"]);
1036                         }
1037                 }
1038
1039
1040                 return $data;
1041         }
1042
1043         /**
1044          * @brief Check for Diaspora contact
1045          *
1046          * @param array $webfinger Webfinger data
1047          *
1048          * @return array Diaspora data
1049          */
1050         private static function diaspora($webfinger) {
1051
1052                 $hcard_url = "";
1053                 $data = array();
1054                 foreach ($webfinger["links"] as $link) {
1055                         if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1056                                 $hcard_url = $link["href"];
1057                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1058                                 $data["baseurl"] = trim($link["href"], '/');
1059                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1060                                 $data["guid"] = $link["href"];
1061                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1062                                 $data["url"] = $link["href"];
1063                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1064                                 $data["poll"] = $link["href"];
1065                         } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1066                                 $data["poco"] = $link["href"];
1067                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1068                                 $data["notify"] = $link["href"];
1069                         } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1070                                 $data["pubkey"] = base64_decode($link["href"]);
1071
1072                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1073                                 if (strstr($data["pubkey"], 'RSA ')) {
1074                                         $data["pubkey"] = rsatopem($data["pubkey"]);
1075                                 }
1076                         }
1077                 }
1078
1079                 if (!isset($data["url"]) || ($hcard_url == "")) {
1080                         return false;
1081                 }
1082
1083                 if (is_array($webfinger["aliases"])) {
1084                         foreach ($webfinger["aliases"] as $alias) {
1085                                 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1086                                         $data["alias"] = $alias;
1087                                 }
1088                         }
1089                 }
1090
1091                 // Fetch further information from the hcard
1092                 $data = self::pollHcard($hcard_url, $data);
1093
1094                 if (!$data) {
1095                         return false;
1096                 }
1097
1098                 if (isset($data["url"])
1099                         && isset($data["guid"])
1100                         && isset($data["baseurl"])
1101                         && isset($data["pubkey"])
1102                         && ($hcard_url != "")
1103                 ) {
1104                         $data["network"] = NETWORK_DIASPORA;
1105
1106                         // The Diaspora handle must always be lowercase
1107                         $data["addr"] = strtolower($data["addr"]);
1108
1109                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1110                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1111                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1112                 } else {
1113                         return false;
1114                 }
1115
1116                 return $data;
1117         }
1118
1119         /**
1120          * @brief Check for OStatus contact
1121          *
1122          * @param array $webfinger Webfinger data
1123          * @param bool $short Short detection mode
1124          *
1125          * @return array|bool OStatus data or "false" on error or "true" on short mode
1126          */
1127         private static function ostatus($webfinger, $short = false) {
1128                 $data = array();
1129                 if (is_array($webfinger["aliases"])) {
1130                         foreach ($webfinger["aliases"] as $alias) {
1131                                 if (strstr($alias, "@")) {
1132                                         $data["addr"] = str_replace('acct:', '', $alias);
1133                                 }
1134                         }
1135                 }
1136
1137                 if (is_string($webfinger["subject"]) && strstr($webfinger["subject"], "@")) {
1138                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1139                 }
1140                 $pubkey = "";
1141                 foreach ($webfinger["links"] as $link) {
1142                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1143                                 && ($link["type"] == "text/html")
1144                                 && ($link["href"] != "")
1145                         ) {
1146                                 $data["url"] = $link["href"];
1147                         } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1148                                 $data["notify"] = $link["href"];
1149                         } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1150                                 $data["poll"] = $link["href"];
1151                         } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1152                                 $pubkey = $link["href"];
1153
1154                                 if (substr($pubkey, 0, 5) === 'data:') {
1155                                         if (strstr($pubkey, ',')) {
1156                                                 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1157                                         } else {
1158                                                 $pubkey = substr($pubkey, 5);
1159                                         }
1160                                 } elseif (normalise_link($pubkey) == 'http://') {
1161                                         $ret = z_fetch_url($pubkey);
1162                                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1163                                                 return false;
1164                                         }
1165                                         $pubkey = $ret['body'];
1166                                 }
1167
1168                                 $key = explode(".", $pubkey);
1169
1170                                 if (sizeof($key) >= 3) {
1171                                         $m = base64url_decode($key[1]);
1172                                         $e = base64url_decode($key[2]);
1173                                         $data["pubkey"] = metopem($m, $e);
1174                                 }
1175                         }
1176                 }
1177
1178                 if (isset($data["notify"]) && isset($data["pubkey"])
1179                         && isset($data["poll"])
1180                         && isset($data["url"])
1181                 ) {
1182                         $data["network"] = NETWORK_OSTATUS;
1183                 } else {
1184                         return false;
1185                 }
1186
1187                 if ($short) {
1188                         return true;
1189                 }
1190
1191                 // Fetch all additional data from the feed
1192                 $ret = z_fetch_url($data["poll"]);
1193                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1194                         return false;
1195                 }
1196                 $feed = $ret['body'];
1197                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1198                 if (!$feed_data) {
1199                         return false;
1200                 }
1201
1202                 if ($feed_data["header"]["author-name"] != "") {
1203                         $data["name"] = $feed_data["header"]["author-name"];
1204                 }
1205                 if ($feed_data["header"]["author-nick"] != "") {
1206                         $data["nick"] = $feed_data["header"]["author-nick"];
1207                 }
1208                 if ($feed_data["header"]["author-avatar"] != "") {
1209                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1210                 }
1211                 if ($feed_data["header"]["author-id"] != "") {
1212                         $data["alias"] = $feed_data["header"]["author-id"];
1213                 }
1214                 if ($feed_data["header"]["author-location"] != "") {
1215                         $data["location"] = $feed_data["header"]["author-location"];
1216                 }
1217                 if ($feed_data["header"]["author-about"] != "") {
1218                         $data["about"] = $feed_data["header"]["author-about"];
1219                 }
1220                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1221                 // So we take the value that we just fetched, although the other one worked as well
1222                 if ($feed_data["header"]["author-link"] != "") {
1223                         $data["url"] = $feed_data["header"]["author-link"];
1224                 }
1225
1226                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1227                         $data['url'] = $data["alias"];
1228                         $data["alias"] = '';
1229                 }
1230
1231                 /// @todo Fetch location and "about" from the feed as well
1232                 return $data;
1233         }
1234
1235         /**
1236          * @brief Fetch data from a pump.io profile page
1237          *
1238          * @param string $profile_link Link to the profile page
1239          *
1240          * @return array profile data
1241          */
1242         private static function pumpioProfileData($profile_link) {
1243
1244                 $doc = new DOMDocument();
1245                 if (!@$doc->loadHTMLFile($profile_link)) {
1246                         return false;
1247                 }
1248
1249                 $xpath = new DomXPath($doc);
1250
1251                 $data = array();
1252
1253                 // This is ugly - but pump.io doesn't seem to know a better way for it
1254                 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1255                 $pos = strpos($data["name"], chr(10));
1256                 if ($pos) {
1257                         $data["name"] = trim(substr($data["name"], 0, $pos));
1258                 }
1259
1260                 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1261                 if ($avatar) {
1262                         foreach ($avatar->attributes as $attribute) {
1263                                 if ($attribute->name == "src") {
1264                                         $data["photo"] = trim($attribute->value);
1265                                 }
1266                         }
1267                 }
1268
1269                 $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
1270                 $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
1271
1272                 return $data;
1273         }
1274
1275         /**
1276          * @brief Check for pump.io contact
1277          *
1278          * @param array $webfinger Webfinger data
1279          *
1280          * @return array pump.io data
1281          */
1282         private static function pumpio($webfinger) {
1283
1284                 $data = array();
1285                 foreach ($webfinger["links"] as $link) {
1286                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1287                                 && ($link["type"] == "text/html")
1288                                 && ($link["href"] != "")
1289                         ) {
1290                                 $data["url"] = $link["href"];
1291                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1292                                 $data["notify"] = $link["href"];
1293                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1294                                 $data["poll"] = $link["href"];
1295                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1296                                 $data["dialback"] = $link["href"];
1297                         }
1298                 }
1299                 if (isset($data["poll"]) && isset($data["notify"])
1300                         && isset($data["dialback"])
1301                         && isset($data["url"])
1302                 ) {
1303                         // by now we use these fields only for the network type detection
1304                         // So we unset all data that isn't used at the moment
1305                         unset($data["dialback"]);
1306
1307                         $data["network"] = NETWORK_PUMPIO;
1308                 } else {
1309                         return false;
1310                 }
1311
1312                 $profile_data = self::pumpioProfileData($data["url"]);
1313
1314                 if (!$profile_data) {
1315                         return false;
1316                 }
1317
1318                 $data = array_merge($data, $profile_data);
1319
1320                 return $data;
1321         }
1322
1323         /**
1324          * @brief Check page for feed link
1325          *
1326          * @param string $url Page link
1327          *
1328          * @return string feed link
1329          */
1330         private static function getFeedLink($url) {
1331                 $doc = new DOMDocument();
1332
1333                 if (!@$doc->loadHTMLFile($url)) {
1334                         return false;
1335                 }
1336
1337                 $xpath = new DomXPath($doc);
1338
1339                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1340                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1341                 if (!is_object($feeds)) {
1342                         return false;
1343                 }
1344
1345                 if ($feeds->length == 0) {
1346                         return false;
1347                 }
1348
1349                 $feed_url = "";
1350
1351                 foreach ($feeds as $feed) {
1352                         $attr = array();
1353                         foreach ($feed->attributes as $attribute) {
1354                                 $attr[$attribute->name] = trim($attribute->value);
1355                         }
1356
1357                         if ($feed_url == "") {
1358                                 $feed_url = $attr["href"];
1359                         }
1360                 }
1361
1362                 return $feed_url;
1363         }
1364
1365         /**
1366          * @brief Check for feed contact
1367          *
1368          * @param string $url Profile link
1369          * @param boolean $probe Do a probe if the page contains a feed link
1370          *
1371          * @return array feed data
1372          */
1373         private static function feed($url, $probe = true) {
1374                 $ret = z_fetch_url($url);
1375                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1376                         return false;
1377                 }
1378                 $feed = $ret['body'];
1379                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1380
1381                 if (!$feed_data) {
1382                         if (!$probe) {
1383                                 return false;
1384                         }
1385
1386                         $feed_url = self::getFeedLink($url);
1387
1388                         if (!$feed_url) {
1389                                 return false;
1390                         }
1391
1392                         return self::feed($feed_url, false);
1393                 }
1394
1395                 if ($feed_data["header"]["author-name"] != "") {
1396                         $data["name"] = $feed_data["header"]["author-name"];
1397                 }
1398
1399                 if ($feed_data["header"]["author-nick"] != "") {
1400                         $data["nick"] = $feed_data["header"]["author-nick"];
1401                 }
1402
1403                 if ($feed_data["header"]["author-avatar"] != "") {
1404                         $data["photo"] = $feed_data["header"]["author-avatar"];
1405                 }
1406
1407                 if ($feed_data["header"]["author-id"] != "") {
1408                         $data["alias"] = $feed_data["header"]["author-id"];
1409                 }
1410
1411                 $data["url"] = $url;
1412                 $data["poll"] = $url;
1413
1414                 if ($feed_data["header"]["author-link"] != "") {
1415                         $data["baseurl"] = $feed_data["header"]["author-link"];
1416                 } else {
1417                         $data["baseurl"] = $data["url"];
1418                 }
1419
1420                 $data["network"] = NETWORK_FEED;
1421
1422                 return $data;
1423         }
1424
1425         /**
1426          * @brief Check for mail contact
1427          *
1428          * @param string $uri Profile link
1429          * @param integer $uid User ID
1430          *
1431          * @return array mail data
1432          */
1433         private static function mail($uri, $uid) {
1434
1435                 if (!validate_email($uri)) {
1436                         return false;
1437                 }
1438
1439                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1440
1441                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1442
1443                 if (dbm::is_result($x) && dbm::is_result($r)) {
1444                         $mailbox = construct_mailbox_name($r[0]);
1445                         $password = '';
1446                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1447                         $mbox = email_connect($mailbox, $r[0]['user'], $password);
1448                         if (!mbox) {
1449                                 return false;
1450                         }
1451                 }
1452
1453                 $msgs = email_poll($mbox, $uri);
1454                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1455
1456                 if (!count($msgs)) {
1457                         return false;
1458                 }
1459
1460                 $phost = substr($uri, strpos($uri, '@') + 1);
1461
1462                 $data = array();
1463                 $data["addr"]    = $uri;
1464                 $data["network"] = NETWORK_MAIL;
1465                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1466                 $data["nick"]    = $data["name"];
1467                 $data["photo"]   = avatar_img($uri);
1468                 $data["url"]     = 'http://'.$phost."/".$data["nick"];
1469                 $data["notify"]  = 'smtp '.random_string();
1470                 $data["poll"]    = 'email '.random_string();
1471
1472                 $x = email_msg_meta($mbox, $msgs[0]);
1473                 if (stristr($x[0]->from, $uri)) {
1474                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1475                 } elseif (stristr($x[0]->to, $uri)) {
1476                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1477                 }
1478                 if (isset($adr)) {
1479                         foreach ($adr as $feadr) {
1480                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1481                                         &&(strcasecmp($feadr->host, $phost) == 0)
1482                                         && (strlen($feadr->personal))
1483                                 ) {
1484                                         $personal = imap_mime_header_decode($feadr->personal);
1485                                         $data["name"] = "";
1486                                         foreach ($personal as $perspart) {
1487                                                 if ($perspart->charset != "default") {
1488                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1489                                                 } else {
1490                                                         $data["name"] .= $perspart->text;
1491                                                 }
1492                                         }
1493
1494                                         $data["name"] = notags($data["name"]);
1495                                 }
1496                         }
1497                 }
1498                 imap_close($mbox);
1499
1500                 return $data;
1501         }
1502
1503         /**
1504          * @brief Mix two paths together to possibly fix missing parts
1505          *
1506          * @param string $avatar Path to the avatar
1507          * @param string $base Another path that is hopefully complete
1508          *
1509          * @return string fixed avatar path
1510          */
1511         public static function fixAvatar($avatar, $base) {
1512                 $base_parts = parse_url($base);
1513
1514                 // Remove all parts that could create a problem
1515                 unset($base_parts['path']);
1516                 unset($base_parts['query']);
1517                 unset($base_parts['fragment']);
1518
1519                 $avatar_parts = parse_url($avatar);
1520
1521                 // Now we mix them
1522                 $parts = array_merge($base_parts, $avatar_parts);
1523
1524                 // And put them together again
1525                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1526                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1527                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1528                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1529                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1530                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1531
1532                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1533
1534                 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
1535
1536                 return $fixed;
1537         }
1538 }