include/Probe.php

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