include/Probe.php

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