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                 // Reset the static variable
  60                 self::$baseurl = '';
  61
  62                 $ssl_url = "https://".$host."/.well-known/host-meta";
  63                 $url = "http://".$host."/.well-known/host-meta";
  64
  65                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
  66                 $redirects = 0;
  67
  68                 $ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
  69                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
  70                         return false;
  71                 }
  72                 $xml = $ret['body'];
  73
  74                 $xrd = parse_xml_string($xml, false);
  75
  76                 if (!is_object($xrd)) {
  77                         $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
  78                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
  79                                 return false;
  80                         }
  81                         $xml = $ret['body'];
  82                         $xrd = parse_xml_string($xml, false);
  83                 }
  84                 if (!is_object($xrd))
  85                         return false;
  86
  87                 $links = xml::element_to_array($xrd);
  88                 if (!isset($links["xrd"]["link"]))
  89                         return false;
  90
  91                 $xrd_data = array();
  92
  93                 foreach ($links["xrd"]["link"] AS $value => $link) {
  94                         if (isset($link["@attributes"]))
  95                                 $attributes = $link["@attributes"];
  96                         elseif ($value == "@attributes")
  97                                 $attributes = $link;
  98                         else
  99                                 continue;
 100
 101                         if (($attributes["rel"] == "lrdd") AND
 102                                 ($attributes["type"] == "application/xrd+xml"))
 103                                 $xrd_data["lrdd-xml"] = $attributes["template"];
 104                         elseif (($attributes["rel"] == "lrdd") AND
 105                                 ($attributes["type"] == "application/json"))
 106                                 $xrd_data["lrdd-json"] = $attributes["template"];
 107                         elseif ($attributes["rel"] == "lrdd")
 108                                 $xrd_data["lrdd"] = $attributes["template"];
 109                 }
 110
 111                 self::$baseurl = "http://".$host;
 112
 113                 return $xrd_data;
 114         }
 115
 116         /**
 117          * @brief Perform Webfinger lookup and return DFRN data
 118          *
 119          * Given an email style address, perform webfinger lookup and
 120          * return the resulting DFRN profile URL, or if no DFRN profile URL
 121          * is located, returns an OStatus subscription template (prefixed
 122          * with the string 'stat:' to identify it as on OStatus template).
 123          * If this isn't an email style address just return $webbie.
 124          * Return an empty string if email-style addresses but webfinger fails,
 125          * or if the resultant personal XRD doesn't contain a supported
 126          * subscription/friend-request attribute.
 127          *
 128          * amended 7/9/2011 to return an hcard which could save potentially loading
 129          * a lengthy content page to scrape dfrn attributes
 130          *
 131          * @param string $webbie Address that should be probed
 132          * @param string $hcard Link to the hcard - is returned by reference
 133          *
 134          * @return string profile link
 135          */
 136
 137         public static function webfinger_dfrn($webbie, &$hcard) {
 138
 139                 $profile_link = '';
 140
 141                 $links = self::lrdd($webbie);
 142                 logger('webfinger_dfrn: '.$webbie.':'.print_r($links,true), LOGGER_DATA);
 143                 if (count($links)) {
 144                         foreach ($links as $link) {
 145                                 if ($link['@attributes']['rel'] === NAMESPACE_DFRN)
 146                                         $profile_link = $link['@attributes']['href'];
 147                                 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) AND ($profile_link == ""))
 148                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 149                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 150                                         $hcard = $link['@attributes']['href'];
 151                         }
 152                 }
 153                 return $profile_link;
 154         }
 155
 156         /**
 157          * @brief Check an URI for LRDD data
 158          *
 159          * this is a replacement for the "lrdd" function in include/network.php.
 160          * It isn't used in this class and has some redundancies in the code.
 161          * When time comes we can check the existing calls for "lrdd" if we can rework them.
 162          *
 163          * @param string $uri Address that should be probed
 164          *
 165          * @return array uri data
 166          */
 167         public static function lrdd($uri) {
 168
 169                 $lrdd = self::xrd($uri);
 170
 171                 if (!$lrdd) {
 172                         $parts = @parse_url($uri);
 173                         if (!$parts)
 174                                 return array();
 175
 176                         $host = $parts["host"];
 177
 178                         $path_parts = explode("/", trim($parts["path"], "/"));
 179
 180                         do {
 181                                 $lrdd = self::xrd($host);
 182                                 $host .= "/".array_shift($path_parts);
 183                         } while (!$lrdd AND (sizeof($path_parts) > 0));
 184                 }
 185
 186                 if (!$lrdd)
 187                         return array();
 188
 189                 foreach ($lrdd AS $key => $link) {
 190                         if ($webfinger)
 191                                 continue;
 192
 193                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
 194                                 continue;
 195
 196                         $path = str_replace('{uri}', urlencode($uri), $link);
 197                         $webfinger = self::webfinger($path);
 198
 199                         if (!$webfinger AND (strstr($uri, "@"))) {
 200                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
 201                                 $webfinger = self::webfinger($path);
 202                         }
 203                 }
 204
 205                 if (!is_array($webfinger["links"]))
 206                         return false;
 207
 208                 $data = array();
 209
 210                 foreach ($webfinger["links"] AS $link)
 211                         $data[] = array("@attributes" => $link);
 212
 213                 if (is_array($webfinger["aliases"]))
 214                         foreach ($webfinger["aliases"] AS $alias)
 215                                 $data[] = array("@attributes" =>
 216                                                         array("rel" => "alias",
 217                                                                 "href" => $alias));
 218
 219                 return $data;
 220         }
 221
 222         /**
 223          * @brief Fetch information (protocol endpoints and user information) about a given uri
 224          *
 225          * @param string $uri Address that should be probed
 226          * @param string $network Test for this specific network
 227          * @param integer $uid User ID for the probe (only used for mails)
 228          * @param boolean $cache Use cached values?
 229          *
 230          * @return array uri data
 231          */
 232         public static function uri($uri, $network = "", $uid = 0, $cache = true) {
 233
 234                 if ($cache) {
 235                         $result = Cache::get("probe_url:".$network.":".$uri);
 236                         if (!is_null($result)) {
 237                                 return $result;
 238                         }
 239                 }
 240
 241                 if ($uid == 0)
 242                         $uid = local_user();
 243
 244                 $data = self::detect($uri, $network, $uid);
 245
 246                 if (!isset($data["url"]))
 247                         $data["url"] = $uri;
 248
 249                 if ($data["photo"] != "")
 250                         $data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
 251                 else
 252                         $data["photo"] = App::get_baseurl().'/images/person-175.jpg';
 253
 254                 if (!isset($data["name"]) OR ($data["name"] == "")) {
 255                         if (isset($data["nick"]))
 256                                 $data["name"] = $data["nick"];
 257
 258                         if ($data["name"] == "")
 259                                 $data["name"] = $data["url"];
 260                 }
 261
 262                 if (!isset($data["nick"]) OR ($data["nick"] == "")) {
 263                         $data["nick"] = strtolower($data["name"]);
 264
 265                         if (strpos($data['nick'], ' '))
 266                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 267                 }
 268
 269                 if (self::$baseurl != "") {
 270                         $data["baseurl"] = self::$baseurl;
 271                 }
 272
 273                 if (!isset($data["network"])) {
 274                         $data["network"] = NETWORK_PHANTOM;
 275                 }
 276
 277                 $data = self::rearrange_data($data);
 278
 279                 // Only store into the cache if the value seems to be valid
 280                 if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
 281                         Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
 282
 283                         /// @todo temporary fix - we need a real contact update function that updates only changing fields
 284                         /// The biggest problem is the avatar picture that could have a reduced image size.
 285                         /// It should only be updated if the existing picture isn't existing anymore.
 286                         if (($data['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
 287                                 $data["name"] AND $data["nick"] AND $data["url"] AND $data["addr"] AND $data["poll"])
 288                                 q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 289                                                 `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 290                                         WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 291                                         dbesc($data["name"]),
 292                                         dbesc($data["nick"]),
 293                                         dbesc($data["url"]),
 294                                         dbesc($data["addr"]),
 295                                         dbesc($data["notify"]),
 296                                         dbesc($data["poll"]),
 297                                         dbesc($data["alias"]),
 298                                         dbesc(datetime_convert()),
 299                                         dbesc(normalise_link($data['url']))
 300                         );
 301                 }
 302
 303                 return $data;
 304         }
 305
 306         /**
 307          * @brief Fetch information (protocol endpoints and user information) about a given uri
 308          *
 309          * This function is only called by the "uri" function that adds caching and rearranging of data.
 310          *
 311          * @param string $uri Address that should be probed
 312          * @param string $network Test for this specific network
 313          * @param integer $uid User ID for the probe (only used for mails)
 314          *
 315          * @return array uri data
 316          */
 317         private function detect($uri, $network, $uid) {
 318                 if (strstr($uri, '@')) {
 319                         // If the URI starts with "mailto:" then jump directly to the mail detection
 320                         if (strpos($url,'mailto:') !== false) {
 321                                 $uri = str_replace('mailto:', '', $url);
 322                                 return self::mail($uri, $uid);
 323                         }
 324
 325                         if ($network == NETWORK_MAIL)
 326                                 return self::mail($uri, $uid);
 327
 328                         // Remove "acct:" from the URI
 329                         $uri = str_replace('acct:', '', $uri);
 330
 331                         $host = substr($uri,strpos($uri, '@') + 1);
 332                         $nick = substr($uri,0, strpos($uri, '@'));
 333
 334                         if (strpos($uri, '@twitter.com'))
 335                                 return array("network" => NETWORK_TWITTER);
 336
 337                         $lrdd = self::xrd($host);
 338
 339                         if (!$lrdd)
 340                                 return self::mail($uri, $uid);
 341
 342                         $addr = $uri;
 343                 } else {
 344                         $parts = parse_url($uri);
 345                         if (!isset($parts["scheme"]) OR
 346                                 !isset($parts["host"]) OR
 347                                 !isset($parts["path"]))
 348                                 return false;
 349
 350                         /// @todo: Ports?
 351                         $host = $parts["host"];
 352
 353                         if ($host == 'twitter.com')
 354                                 return array("network" => NETWORK_TWITTER);
 355
 356                         $lrdd = self::xrd($host);
 357
 358                         $path_parts = explode("/", trim($parts["path"], "/"));
 359
 360                         while (!$lrdd AND (sizeof($path_parts) > 1)) {
 361                                 $host .= "/".array_shift($path_parts);
 362                                 $lrdd = self::xrd($host);
 363                         }
 364                         if (!$lrdd)
 365                                 return self::feed($uri);
 366
 367                         $nick = array_pop($path_parts);
 368                         $addr = $nick."@".$host;
 369                 }
 370                 $webfinger = false;
 371
 372                 /// @todo Do we need the prefix "acct:" or "acct://"?
 373
 374                 foreach ($lrdd AS $key => $link) {
 375                         if ($webfinger)
 376                                 continue;
 377
 378                         if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
 379                                 continue;
 380
 381                         // Try webfinger with the address (user@domain.tld)
 382                         $path = str_replace('{uri}', urlencode($addr), $link);
 383                         $webfinger = self::webfinger($path);
 384
 385                         // Mastodon needs to have it with "acct:"
 386                         if (!$webfinger) {
 387                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $link);
 388                                 $webfinger = self::webfinger($path);
 389                         }
 390
 391                         // If webfinger wasn't successful then try it with the URL - possibly in the format https://...
 392                         if (!$webfinger AND ($uri != $addr)) {
 393                                 $path = str_replace('{uri}', urlencode($uri), $link);
 394                                 $webfinger = self::webfinger($path);
 395
 396                                 // Since the detection with the address wasn't successful, we delete it.
 397                                 if ($webfinger) {
 398                                         $nick = "";
 399                                         $addr = "";
 400                                 }
 401                         }
 402
 403                 }
 404                 if (!$webfinger)
 405                         return self::feed($uri);
 406
 407                 $result = false;
 408
 409                 logger("Probing ".$uri, LOGGER_DEBUG);
 410
 411                 if (in_array($network, array("", NETWORK_DFRN)))
 412                         $result = self::dfrn($webfinger);
 413                 if ((!$result AND ($network == "")) OR ($network == NETWORK_DIASPORA))
 414                         $result = self::diaspora($webfinger);
 415                 if ((!$result AND ($network == "")) OR ($network == NETWORK_OSTATUS))
 416                         $result = self::ostatus($webfinger);
 417                 if ((!$result AND ($network == "")) OR ($network == NETWORK_PUMPIO))
 418                         $result = self::pumpio($webfinger);
 419                 if ((!$result AND ($network == "")) OR ($network == NETWORK_FEED))
 420                         $result = self::feed($uri);
 421                 else {
 422                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 423                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 424                         if ((!isset($result["nick"]) OR ($result["nick"] == "") OR (strstr($result["nick"], " "))) AND ($nick != ""))
 425                                 $result["nick"] = $nick;
 426
 427                         if ((!isset($result["addr"]) OR ($result["addr"] == "")) AND ($addr != ""))
 428                                 $result["addr"] = $addr;
 429                 }
 430
 431                 logger($uri." is ".$result["network"], LOGGER_DEBUG);
 432
 433                 if (!isset($result["baseurl"]) OR ($result["baseurl"] == "")) {
 434                         $pos = strpos($result["url"], $host);
 435                         if ($pos)
 436                                 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
 437                 }
 438
 439                 return $result;
 440         }
 441
 442         /**
 443          * @brief Perform a webfinger request.
 444          *
 445          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 446          *
 447          * @param string $url Address that should be probed
 448          *
 449          * @return array webfinger data
 450          */
 451         private function webfinger($url) {
 452
 453                 $xrd_timeout = Config::get('system','xrd_timeout', 20);
 454                 $redirects = 0;
 455
 456                 $ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
 457                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 458                         return false;
 459                 }
 460                 $data = $ret['body'];
 461
 462                 $xrd = parse_xml_string($data, false);
 463
 464                 if (!is_object($xrd)) {
 465                         // If it is not XML, maybe it is JSON
 466                         $webfinger = json_decode($data, true);
 467
 468                         if (!isset($webfinger["links"]))
 469                                 return false;
 470
 471                         return $webfinger;
 472                 }
 473
 474                 $xrd_arr = xml::element_to_array($xrd);
 475                 if (!isset($xrd_arr["xrd"]["link"]))
 476                         return false;
 477
 478                 $webfinger = array();
 479
 480                 if (isset($xrd_arr["xrd"]["subject"]))
 481                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 482
 483                 if (isset($xrd_arr["xrd"]["alias"]))
 484                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 485
 486                 $webfinger["links"] = array();
 487
 488                 foreach ($xrd_arr["xrd"]["link"] AS $value => $data) {
 489                         if (isset($data["@attributes"]))
 490                                 $attributes = $data["@attributes"];
 491                         elseif ($value == "@attributes")
 492                                 $attributes = $data;
 493                         else
 494                                 continue;
 495
 496                         $webfinger["links"][] = $attributes;
 497                 }
 498                 return $webfinger;
 499         }
 500
 501         /**
 502          * @brief Poll the Friendica specific noscrape page.
 503          *
 504          * "noscrape" is a faster alternative to fetch the data from the hcard.
 505          * This functionality was originally created for the directory.
 506          *
 507          * @param string $noscrape Link to the noscrape page
 508          * @param array $data The already fetched data
 509          *
 510          * @return array noscrape data
 511          */
 512         private function poll_noscrape($noscrape, $data) {
 513                 $ret = z_fetch_url($noscrape);
 514                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 515                         return false;
 516                 }
 517                 $content = $ret['body'];
 518                 if (!$content) {
 519                         return false;
 520                 }
 521
 522                 $json = json_decode($content, true);
 523                 if (!is_array($json))
 524                         return false;
 525
 526                 if (isset($json["fn"]))
 527                         $data["name"] = $json["fn"];
 528
 529                 if (isset($json["addr"]))
 530                         $data["addr"] = $json["addr"];
 531
 532                 if (isset($json["nick"]))
 533                         $data["nick"] = $json["nick"];
 534
 535                 if (isset($json["comm"]))
 536                         $data["community"] = $json["comm"];
 537
 538                 if (isset($json["tags"])) {
 539                         $keywords = implode(" ", $json["tags"]);
 540                         if ($keywords != "")
 541                                 $data["keywords"] = $keywords;
 542                 }
 543
 544                 $location = formatted_location($json);
 545                 if ($location)
 546                         $data["location"] = $location;
 547
 548                 if (isset($json["about"]))
 549                         $data["about"] = $json["about"];
 550
 551                 if (isset($json["key"]))
 552                         $data["pubkey"] = $json["key"];
 553
 554                 if (isset($json["photo"]))
 555                         $data["photo"] = $json["photo"];
 556
 557                 if (isset($json["dfrn-request"]))
 558                         $data["request"] = $json["dfrn-request"];
 559
 560                 if (isset($json["dfrn-confirm"]))
 561                         $data["confirm"] = $json["dfrn-confirm"];
 562
 563                 if (isset($json["dfrn-notify"]))
 564                         $data["notify"] = $json["dfrn-notify"];
 565
 566                 if (isset($json["dfrn-poll"]))
 567                         $data["poll"] = $json["dfrn-poll"];
 568
 569                 return $data;
 570         }
 571
 572         /**
 573          * @brief Check for valid DFRN data
 574          *
 575          * @param array $data DFRN data
 576          *
 577          * @return int Number of errors
 578          */
 579         public static function valid_dfrn($data) {
 580                 $errors = 0;
 581                 if(!isset($data['key']))
 582                         $errors ++;
 583                 if(!isset($data['dfrn-request']))
 584                         $errors ++;
 585                 if(!isset($data['dfrn-confirm']))
 586                         $errors ++;
 587                 if(!isset($data['dfrn-notify']))
 588                         $errors ++;
 589                 if(!isset($data['dfrn-poll']))
 590                         $errors ++;
 591                 return $errors;
 592         }
 593
 594         /**
 595          * @brief Fetch data from a DFRN profile page and via "noscrape"
 596          *
 597          * @param string $profile Link to the profile page
 598          *
 599          * @return array profile data
 600          */
 601         public static function profile($profile) {
 602
 603                 $data = array();
 604
 605                 logger("Check profile ".$profile, LOGGER_DEBUG);
 606
 607                 // Fetch data via noscrape - this is faster
 608                 $noscrape = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile);
 609                 $data = self::poll_noscrape($noscrape, $data);
 610
 611                 if (!isset($data["notify"]) OR !isset($data["confirm"]) OR
 612                         !isset($data["request"]) OR !isset($data["poll"]) OR
 613                         !isset($data["poco"]) OR !isset($data["name"]) OR
 614                         !isset($data["photo"]))
 615                         $data = self::poll_hcard($profile, $data, true);
 616
 617                 $prof_data = array();
 618                 $prof_data["addr"] = $data["addr"];
 619                 $prof_data["nick"] = $data["nick"];
 620                 $prof_data["dfrn-request"] = $data["request"];
 621                 $prof_data["dfrn-confirm"] = $data["confirm"];
 622                 $prof_data["dfrn-notify"] = $data["notify"];
 623                 $prof_data["dfrn-poll"] = $data["poll"];
 624                 $prof_data["dfrn-poco"] = $data["poco"];
 625                 $prof_data["photo"] = $data["photo"];
 626                 $prof_data["fn"] = $data["name"];
 627                 $prof_data["key"] = $data["pubkey"];
 628
 629                 logger("Result for profile ".$profile.": ".print_r($prof_data, true), LOGGER_DEBUG);
 630
 631                 return $prof_data;
 632         }
 633
 634         /**
 635          * @brief Check for DFRN contact
 636          *
 637          * @param array $webfinger Webfinger data
 638          *
 639          * @return array DFRN data
 640          */
 641         private function dfrn($webfinger) {
 642
 643                 $hcard = "";
 644                 $data = array();
 645                 foreach ($webfinger["links"] AS $link) {
 646                         if (($link["rel"] == NAMESPACE_DFRN) AND ($link["href"] != ""))
 647                                 $data["network"] = NETWORK_DFRN;
 648                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 649                                 $data["poll"] = $link["href"];
 650                         elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 651                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 652                                 $data["url"] = $link["href"];
 653                         elseif (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 654                                 $hcard = $link["href"];
 655                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 656                                 $data["poco"] = $link["href"];
 657                         elseif (($link["rel"] == "http://webfinger.net/rel/avatar") AND ($link["href"] != ""))
 658                                 $data["photo"] = $link["href"];
 659
 660                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 661                                 $data["baseurl"] = trim($link["href"], '/');
 662                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 663                                 $data["guid"] = $link["href"];
 664                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 665                                 $data["pubkey"] = base64_decode($link["href"]);
 666
 667                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 668                                 if (strstr($data["pubkey"], 'RSA '))
 669                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 670                         }
 671                 }
 672
 673                 if (!isset($data["network"]) OR ($hcard == ""))
 674                         return false;
 675
 676                 // Fetch data via noscrape - this is faster
 677                 $noscrape = str_replace("/hcard/", "/noscrape/", $hcard);
 678                 $data = self::poll_noscrape($noscrape, $data);
 679
 680                 if (isset($data["notify"]) AND isset($data["confirm"]) AND isset($data["request"]) AND
 681                         isset($data["poll"]) AND isset($data["name"]) AND isset($data["photo"]))
 682                         return $data;
 683
 684                 $data = self::poll_hcard($hcard, $data, true);
 685
 686                 return $data;
 687         }
 688
 689         /**
 690          * @brief Poll the hcard page (Diaspora and Friendica specific)
 691          *
 692          * @param string $hcard Link to the hcard page
 693          * @param array $data The already fetched data
 694          * @param boolean $dfrn Poll DFRN specific data
 695          *
 696          * @return array hcard data
 697          */
 698         private function poll_hcard($hcard, $data, $dfrn = false) {
 699                 $ret = z_fetch_url($hcard);
 700                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 701                         return false;
 702                 }
 703                 $content = $ret['body'];
 704                 if (!$content) {
 705                         return false;
 706                 }
 707
 708                 $doc = new DOMDocument();
 709                 if (!@$doc->loadHTML($content))
 710                         return false;
 711
 712                 $xpath = new DomXPath($doc);
 713
 714                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
 715                 if (!is_object($vcards))
 716                         return false;
 717
 718                 if ($vcards->length > 0) {
 719                         $vcard = $vcards->item(0);
 720
 721                         // We have to discard the guid from the hcard in favour of the guid from lrdd
 722                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
 723                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
 724                         if (($search->length > 0) AND ($data["guid"] == ""))
 725                                 $data["guid"] = $search->item(0)->nodeValue;
 726
 727                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
 728                         if ($search->length > 0)
 729                                 $data["nick"] = $search->item(0)->nodeValue;
 730
 731                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
 732                         if ($search->length > 0)
 733                                 $data["name"] = $search->item(0)->nodeValue;
 734
 735                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
 736                         if ($search->length > 0)
 737                                 $data["searchable"] = $search->item(0)->nodeValue;
 738
 739                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
 740                         if ($search->length > 0) {
 741                                 $data["pubkey"] = $search->item(0)->nodeValue;
 742                                 if (strstr($data["pubkey"], 'RSA '))
 743                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 744                         }
 745
 746                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
 747                         if ($search->length > 0)
 748                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
 749                 }
 750
 751                 $avatar = array();
 752                 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
 753                 foreach ($photos AS $photo) {
 754                         $attr = array();
 755                         foreach ($photo->attributes as $attribute) {
 756                                 $attr[$attribute->name] = trim($attribute->value);
 757                         }
 758
 759                         if (isset($attr["src"]) AND isset($attr["width"])) {
 760                                 $avatar[$attr["width"]] = $attr["src"];
 761                         }
 762
 763                         // We don't have a width. So we just take everything that we got.
 764                         // This is a Hubzilla workaround which doesn't send a width.
 765                         if ((sizeof($avatar) == 0) AND isset($attr["src"])) {
 766                                 $avatar[] = $attr["src"];
 767                         }
 768                 }
 769
 770                 if (sizeof($avatar)) {
 771                         ksort($avatar);
 772                         $data["photo"] = array_pop($avatar);
 773                 }
 774
 775                 if ($dfrn) {
 776                         // Poll DFRN specific data
 777                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
 778                         if ($search->length > 0) {
 779                                 foreach ($search AS $link) {
 780                                         //$data["request"] = $search->item(0)->nodeValue;
 781                                         $attr = array();
 782                                         foreach ($link->attributes as $attribute)
 783                                                 $attr[$attribute->name] = trim($attribute->value);
 784
 785                                         $data[substr($attr["rel"], 5)] = $attr["href"];
 786                                 }
 787                         }
 788
 789                         // Older Friendica versions had used the "uid" field differently than newer versions
 790                         if ($data["nick"] == $data["guid"])
 791                                 unset($data["guid"]);
 792                 }
 793
 794
 795                 return $data;
 796         }
 797
 798         /**
 799          * @brief Check for Diaspora contact
 800          *
 801          * @param array $webfinger Webfinger data
 802          *
 803          * @return array Diaspora data
 804          */
 805         private function diaspora($webfinger) {
 806
 807                 $hcard = "";
 808                 $data = array();
 809                 foreach ($webfinger["links"] AS $link) {
 810                         if (($link["rel"] == "http://microformats.org/profile/hcard") AND ($link["href"] != ""))
 811                                 $hcard = $link["href"];
 812                         elseif (($link["rel"] == "http://joindiaspora.com/seed_location") AND ($link["href"] != ""))
 813                                 $data["baseurl"] = trim($link["href"], '/');
 814                         elseif (($link["rel"] == "http://joindiaspora.com/guid") AND ($link["href"] != ""))
 815                                 $data["guid"] = $link["href"];
 816                         elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 817                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 818                                 $data["url"] = $link["href"];
 819                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 820                                 $data["poll"] = $link["href"];
 821                         elseif (($link["rel"] == NAMESPACE_POCO) AND ($link["href"] != ""))
 822                                 $data["poco"] = $link["href"];
 823                         elseif (($link["rel"] == "salmon") AND ($link["href"] != ""))
 824                                 $data["notify"] = $link["href"];
 825                         elseif (($link["rel"] == "diaspora-public-key") AND ($link["href"] != "")) {
 826                                 $data["pubkey"] = base64_decode($link["href"]);
 827
 828                                 //if (strstr($data["pubkey"], 'RSA ') OR ($link["type"] == "RSA"))
 829                                 if (strstr($data["pubkey"], 'RSA '))
 830                                         $data["pubkey"] = rsatopem($data["pubkey"]);
 831                         }
 832                 }
 833
 834                 if (!isset($data["url"]) OR ($hcard == ""))
 835                         return false;
 836
 837                 if (is_array($webfinger["aliases"]))
 838                         foreach ($webfinger["aliases"] AS $alias)
 839                                 if (normalise_link($alias) != normalise_link($data["url"]) AND !strstr($alias, "@"))
 840                                         $data["alias"] = $alias;
 841
 842                 // Fetch further information from the hcard
 843                 $data = self::poll_hcard($hcard, $data);
 844
 845                 if (!$data)
 846                         return false;
 847
 848                 if (isset($data["url"]) AND isset($data["guid"]) AND isset($data["baseurl"]) AND
 849                         isset($data["pubkey"]) AND ($hcard != "")) {
 850                         $data["network"] = NETWORK_DIASPORA;
 851
 852                         // The Diaspora handle must always be lowercase
 853                         $data["addr"] = strtolower($data["addr"]);
 854
 855                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
 856                         $data["notify"] = $data["baseurl"]."/receive/users/".$data["guid"];
 857                         $data["batch"] = $data["baseurl"]."/receive/public";
 858                 } else
 859                         return false;
 860
 861                 return $data;
 862         }
 863
 864         /**
 865          * @brief Check for OStatus contact
 866          *
 867          * @param array $webfinger Webfinger data
 868          *
 869          * @return array OStatus data
 870          */
 871         private function ostatus($webfinger) {
 872
 873                 $data = array();
 874                 if (is_array($webfinger["aliases"]))
 875                         foreach($webfinger["aliases"] AS $alias)
 876                                 if (strstr($alias, "@"))
 877                                         $data["addr"] = str_replace('acct:', '', $alias);
 878
 879                 if (is_string($webfinger["subject"]) AND strstr($webfinger["subject"], "@"))
 880                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
 881
 882                 $pubkey = "";
 883                 foreach ($webfinger["links"] AS $link) {
 884                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
 885                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
 886                                 $data["url"] = $link["href"];
 887                         elseif (($link["rel"] == "salmon") AND ($link["href"] != ""))
 888                                 $data["notify"] = $link["href"];
 889                         elseif (($link["rel"] == NAMESPACE_FEED) AND ($link["href"] != ""))
 890                                 $data["poll"] = $link["href"];
 891                         elseif (($link["rel"] == "magic-public-key") AND ($link["href"] != "")) {
 892                                 $pubkey = $link["href"];
 893
 894                                 if (substr($pubkey, 0, 5) === 'data:') {
 895                                         if (strstr($pubkey, ','))
 896                                                 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
 897                                         else
 898                                                 $pubkey = substr($pubkey, 5);
 899                                 } elseif (normalise_link($pubkey) == 'http://') {
 900                                         $ret = z_fetch_url($pubkey);
 901                                         if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 902                                                 return false;
 903                                         }
 904                                         $pubkey = $ret['body'];
 905                                 }
 906
 907                                 $key = explode(".", $pubkey);
 908
 909                                 if (sizeof($key) >= 3) {
 910                                         $m = base64url_decode($key[1]);
 911                                         $e = base64url_decode($key[2]);
 912                                         $data["pubkey"] = metopem($m,$e);
 913                                 }
 914
 915                         }
 916                 }
 917
 918                 if (isset($data["notify"]) AND isset($data["pubkey"]) AND
 919                         isset($data["poll"]) AND isset($data["url"])) {
 920                         $data["network"] = NETWORK_OSTATUS;
 921                 } else
 922                         return false;
 923
 924                 // Fetch all additional data from the feed
 925                 $ret = z_fetch_url($data["poll"]);
 926                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
 927                         return false;
 928                 }
 929                 $feed = $ret['body'];
 930                 $feed_data = feed_import($feed,$dummy1,$dummy2, $dummy3, true);
 931                 if (!$feed_data)
 932                         return false;
 933
 934                 if ($feed_data["header"]["author-name"] != "")
 935                         $data["name"] = $feed_data["header"]["author-name"];
 936
 937                 if ($feed_data["header"]["author-nick"] != "")
 938                         $data["nick"] = $feed_data["header"]["author-nick"];
 939
 940                 if ($feed_data["header"]["author-avatar"] != "")
 941                         $data["photo"] = $feed_data["header"]["author-avatar"];
 942
 943                 if ($feed_data["header"]["author-id"] != "")
 944                         $data["alias"] = $feed_data["header"]["author-id"];
 945
 946                 if ($feed_data["header"]["author-location"] != "")
 947                         $data["location"] = $feed_data["header"]["author-location"];
 948
 949                 if ($feed_data["header"]["author-about"] != "")
 950                         $data["about"] = $feed_data["header"]["author-about"];
 951
 952                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
 953                 // So we take the value that we just fetched, although the other one worked as well
 954                 if ($feed_data["header"]["author-link"] != "")
 955                         $data["url"] = $feed_data["header"]["author-link"];
 956
 957                 /// @todo Fetch location and "about" from the feed as well
 958                 return $data;
 959         }
 960
 961         /**
 962          * @brief Fetch data from a pump.io profile page
 963          *
 964          * @param string $profile Link to the profile page
 965          *
 966          * @return array profile data
 967          */
 968         private function pumpio_profile_data($profile) {
 969
 970                 $doc = new DOMDocument();
 971                 if (!@$doc->loadHTMLFile($profile))
 972                         return false;
 973
 974                 $xpath = new DomXPath($doc);
 975
 976                 $data = array();
 977
 978                 // This is ugly - but pump.io doesn't seem to know a better way for it
 979                 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
 980                 $pos = strpos($data["name"], chr(10));
 981                 if ($pos)
 982                         $data["name"] = trim(substr($data["name"], 0, $pos));
 983
 984                 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
 985                 if ($avatar)
 986                         foreach ($avatar->attributes as $attribute)
 987                                 if ($attribute->name == "src")
 988                                         $data["photo"] = trim($attribute->value);
 989
 990                 $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
 991                 $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
 992
 993                 return $data;
 994         }
 995
 996         /**
 997          * @brief Check for pump.io contact
 998          *
 999          * @param array $webfinger Webfinger data
1000          *
1001          * @return array pump.io data
1002          */
1003         private function pumpio($webfinger) {
1004
1005                 $data = array();
1006                 foreach ($webfinger["links"] AS $link) {
1007                         if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
1008                                 ($link["type"] == "text/html") AND ($link["href"] != ""))
1009                                 $data["url"] = $link["href"];
1010                         elseif (($link["rel"] == "activity-inbox") AND ($link["href"] != ""))
1011                                 $data["notify"] = $link["href"];
1012                         elseif (($link["rel"] == "activity-outbox") AND ($link["href"] != ""))
1013                                 $data["poll"] = $link["href"];
1014                         elseif (($link["rel"] == "dialback") AND ($link["href"] != ""))
1015                                 $data["dialback"] = $link["href"];
1016                 }
1017                 if (isset($data["poll"]) AND isset($data["notify"]) AND
1018                         isset($data["dialback"]) AND isset($data["url"])) {
1019
1020                         // by now we use these fields only for the network type detection
1021                         // So we unset all data that isn't used at the moment
1022                         unset($data["dialback"]);
1023
1024                         $data["network"] = NETWORK_PUMPIO;
1025                 } else
1026                         return false;
1027
1028                 $profile_data = self::pumpio_profile_data($data["url"]);
1029
1030                 if (!$profile_data)
1031                         return false;
1032
1033                 $data = array_merge($data, $profile_data);
1034
1035                 return $data;
1036         }
1037
1038         /**
1039          * @brief Check page for feed link
1040          *
1041          * @param string $url Page link
1042          *
1043          * @return string feed link
1044          */
1045         private function get_feed_link($url) {
1046                 $doc = new DOMDocument();
1047
1048                 if (!@$doc->loadHTMLFile($url))
1049                         return false;
1050
1051                 $xpath = new DomXPath($doc);
1052
1053                 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1054                 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1055                 if (!is_object($feeds))
1056                         return false;
1057
1058                 if ($feeds->length == 0)
1059                         return false;
1060
1061                 $feed_url = "";
1062
1063                 foreach ($feeds AS $feed) {
1064                         $attr = array();
1065                         foreach ($feed->attributes as $attribute)
1066                         $attr[$attribute->name] = trim($attribute->value);
1067
1068                         if ($feed_url == "")
1069                                 $feed_url = $attr["href"];
1070                 }
1071
1072                 return $feed_url;
1073         }
1074
1075         /**
1076          * @brief Check for feed contact
1077          *
1078          * @param string $url Profile link
1079          * @param boolean $probe Do a probe if the page contains a feed link
1080          *
1081          * @return array feed data
1082          */
1083         private function feed($url, $probe = true) {
1084                 $ret = z_fetch_url($url);
1085                 if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
1086                         return false;
1087                 }
1088                 $feed = $ret['body'];
1089                 $feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
1090
1091                 if (!$feed_data) {
1092                         if (!$probe)
1093                                 return false;
1094
1095                         $feed_url = self::get_feed_link($url);
1096
1097                         if (!$feed_url)
1098                                 return false;
1099
1100                         return self::feed($feed_url, false);
1101                 }
1102
1103                 if ($feed_data["header"]["author-name"] != "")
1104                         $data["name"] = $feed_data["header"]["author-name"];
1105
1106                 if ($feed_data["header"]["author-nick"] != "")
1107                         $data["nick"] = $feed_data["header"]["author-nick"];
1108
1109                 if ($feed_data["header"]["author-avatar"] != "")
1110                         $data["photo"] = $feed_data["header"]["author-avatar"];
1111
1112                 if ($feed_data["header"]["author-id"] != "")
1113                         $data["alias"] = $feed_data["header"]["author-id"];
1114
1115                 $data["url"] = $url;
1116                 $data["poll"] = $url;
1117
1118                 if ($feed_data["header"]["author-link"] != "")
1119                         $data["baseurl"] = $feed_data["header"]["author-link"];
1120                 else
1121                         $data["baseurl"] = $data["url"];
1122
1123                 $data["network"] = NETWORK_FEED;
1124
1125                 return $data;
1126         }
1127
1128         /**
1129          * @brief Check for mail contact
1130          *
1131          * @param string $uri Profile link
1132          * @param integer $uid User ID
1133          *
1134          * @return array mail data
1135          */
1136         private function mail($uri, $uid) {
1137
1138                 if (!validate_email($uri))
1139                         return false;
1140
1141                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1142
1143                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1144
1145                 if (dbm::is_result($x) && dbm::is_result($r)) {
1146                         $mailbox = construct_mailbox_name($r[0]);
1147                         $password = '';
1148                         openssl_private_decrypt(hex2bin($r[0]['pass']), $password,$x[0]['prvkey']);
1149                         $mbox = email_connect($mailbox,$r[0]['user'], $password);
1150                         if(!mbox)
1151                                 return false;
1152                 }
1153
1154                 $msgs = email_poll($mbox, $uri);
1155                 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1156
1157                 if (!count($msgs))
1158                         return false;
1159
1160                 $data = array();
1161
1162                 $data["addr"] = $uri;
1163                 $data["network"] = NETWORK_MAIL;
1164                 $data["name"] = substr($uri, 0, strpos($uri,'@'));
1165                 $data["nick"] = $data["name"];
1166                 $data["photo"] = avatar_img($uri);
1167
1168                 $phost = substr($uri, strpos($uri,'@') + 1);
1169                 $data["url"] = 'http://'.$phost."/".$data["nick"];
1170                 $data["notify"] = 'smtp '.random_string();
1171                 $data["poll"] = 'email '.random_string();
1172
1173                 $x = email_msg_meta($mbox, $msgs[0]);
1174                 if(stristr($x[0]->from, $uri))
1175                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1176                 elseif(stristr($x[0]->to, $uri))
1177                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1178                 if(isset($adr)) {
1179                         foreach($adr as $feadr) {
1180                                 if((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1181                                         &&(strcasecmp($feadr->host, $phost) == 0)
1182                                         && (strlen($feadr->personal))) {
1183
1184                                         $personal = imap_mime_header_decode($feadr->personal);
1185                                         $data["name"] = "";
1186                                         foreach($personal as $perspart)
1187                                                 if ($perspart->charset != "default")
1188                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1189                                                 else
1190                                                         $data["name"] .= $perspart->text;
1191
1192                                         $data["name"] = notags($data["name"]);
1193                                 }
1194                         }
1195                 }
1196                 imap_close($mbox);
1197
1198                 return $data;
1199         }
1200 }
1201 ?>