src/Util/Network.php

   1 <?php
   2 /**
   3  * @file src/Util/Network.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use Friendica\App;
   8 use Friendica\Core\Addon;
   9 use Friendica\Core\L10n;
  10 use Friendica\Core\System;
  11 use Friendica\Core\Config;
  12 use Friendica\Network\Probe;
  13 use Friendica\Object\Image;
  14 use Friendica\Util\XML;
  15
  16 require_once 'library/slinky.php';
  17
  18 class Network
  19 {
  20         /**
  21          * @brief Curl wrapper
  22          *
  23          * If binary flag is true, return binary results.
  24          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
  25          * to preserve cookies from one request to the next.
  26          *
  27          * @param string  $url            URL to fetch
  28          * @param boolean $binary         default false
  29          *                                TRUE if asked to return binary results (file download)
  30          * @param integer $redirects      The recursion counter for internal use - default 0
  31          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  32          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  33          * @param string  $cookiejar      Path to cookie jar file
  34          *
  35          * @return string The fetched content
  36          */
  37         public static function fetchURL($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  38         {
  39                 $ret = self::zFetchURL(
  40                         $url,
  41                         $binary,
  42                         $redirects,
  43                         ['timeout'=>$timeout,
  44                         'accept_content'=>$accept_content,
  45                         'cookiejar'=>$cookiejar
  46                         ]
  47                 );
  48
  49                 return($ret['body']);
  50         }
  51
  52         /**
  53          * @brief fetches an URL.
  54          *
  55          * @param string  $url       URL to fetch
  56          * @param boolean $binary    default false
  57          *                           TRUE if asked to return binary results (file download)
  58          * @param int     $redirects The recursion counter for internal use - default 0
  59          * @param array   $opts      (optional parameters) assoziative array with:
  60          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  61          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  62          *                           'http_auth' => username:password
  63          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  64          *                           'nobody' => only return the header
  65          *                           'cookiejar' => path to cookie jar file
  66          *
  67          * @return array an assoziative array with:
  68          *    int 'return_code' => HTTP return code or 0 if timeout or failure
  69          *    boolean 'success' => boolean true (if HTTP 2xx result) or false
  70          *    string 'redirect_url' => in case of redirect, content was finally retrieved from this URL
  71          *    string 'header' => HTTP headers
  72          *    string 'body' => fetched content
  73          */
  74         public static function zFetchURL($url, $binary = false, &$redirects = 0, $opts = [])
  75         {
  76                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
  77
  78                 $stamp1 = microtime(true);
  79
  80                 $a = get_app();
  81
  82                 if (blocked_url($url)) {
  83                         logger('z_fetch_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
  84                         return $ret;
  85                 }
  86
  87                 $ch = @curl_init($url);
  88
  89                 if (($redirects > 8) || (!$ch)) {
  90                         return $ret;
  91                 }
  92
  93                 @curl_setopt($ch, CURLOPT_HEADER, true);
  94
  95                 if (x($opts, "cookiejar")) {
  96                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
  97                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
  98                 }
  99
 100                 // These settings aren't needed. We're following the location already.
 101                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 102                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 103
 104                 if (x($opts, 'accept_content')) {
 105                         curl_setopt(
 106                                 $ch,
 107                                 CURLOPT_HTTPHEADER,
 108                                 ['Accept: ' . $opts['accept_content']]
 109                         );
 110                 }
 111
 112                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 113                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 114
 115                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 116
 117                 if ($range > 0) {
 118                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 119                 }
 120
 121                 // Without this setting it seems as if some webservers send compressed content
 122                 // This seems to confuse curl so that it shows this uncompressed.
 123                 /// @todo  We could possibly set this value to "gzip" or something similar
 124                 curl_setopt($ch, CURLOPT_ENCODING, '');
 125
 126                 if (x($opts, 'headers')) {
 127                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 128                 }
 129
 130                 if (x($opts, 'nobody')) {
 131                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 132                 }
 133
 134                 if (x($opts, 'timeout')) {
 135                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 136                 } else {
 137                         $curl_time = Config::get('system', 'curl_timeout', 60);
 138                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 139                 }
 140
 141                 // by default we will allow self-signed certs
 142                 // but you can override this
 143
 144                 $check_cert = Config::get('system', 'verifyssl');
 145                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 146
 147                 if ($check_cert) {
 148                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 149                 }
 150
 151                 $proxy = Config::get('system', 'proxy');
 152
 153                 if (strlen($proxy)) {
 154                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 155                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 156                         $proxyuser = @Config::get('system', 'proxyuser');
 157
 158                         if (strlen($proxyuser)) {
 159                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 160                         }
 161                 }
 162
 163                 if (Config::get('system', 'ipv4_resolve', false)) {
 164                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 165                 }
 166
 167                 if ($binary) {
 168                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 169                 }
 170
 171                 $a->set_curl_code(0);
 172
 173                 // don't let curl abort the entire application
 174                 // if it throws any errors.
 175
 176                 $s = @curl_exec($ch);
 177                 $curl_info = @curl_getinfo($ch);
 178
 179                 // Special treatment for HTTP Code 416
 180                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 181                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 182                         @curl_setopt($ch, CURLOPT_RANGE, '');
 183                         $s = @curl_exec($ch);
 184                         $curl_info = @curl_getinfo($ch);
 185                 }
 186
 187                 if (curl_errno($ch) !== CURLE_OK) {
 188                         logger('fetch_url error fetching ' . $url . ': ' . curl_error($ch), LOGGER_NORMAL);
 189                 }
 190
 191                 $ret['errno'] = curl_errno($ch);
 192
 193                 $base = $s;
 194                 $ret['info'] = $curl_info;
 195
 196                 $http_code = $curl_info['http_code'];
 197
 198                 logger('fetch_url ' . $url . ': ' . $http_code . " " . $s, LOGGER_DATA);
 199                 $header = '';
 200
 201                 // Pull out multiple headers, e.g. proxy and continuation headers
 202                 // allow for HTTP/2.x without fixing code
 203
 204                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 205                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 206                         $header .= $chunk;
 207                         $base = substr($base, strlen($chunk));
 208                 }
 209
 210                 $a->set_curl_code($http_code);
 211                 $a->set_curl_content_type($curl_info['content_type']);
 212                 $a->set_curl_headers($header);
 213
 214                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 215                         $new_location_info = @parse_url($curl_info['redirect_url']);
 216                         $old_location_info = @parse_url($curl_info['url']);
 217
 218                         $newurl = $curl_info['redirect_url'];
 219
 220                         if (($new_location_info['path'] == '') && ( $new_location_info['host'] != '')) {
 221                                 $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
 222                         }
 223
 224                         $matches = [];
 225
 226                         if (preg_match('/(Location:|URI:)(.*?)\n/i', $header, $matches)) {
 227                                 $newurl = trim(array_pop($matches));
 228                         }
 229                         if (strpos($newurl, '/') === 0) {
 230                                 $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
 231                         }
 232
 233                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 234                                 $redirects++;
 235                                 @curl_close($ch);
 236                                 return self::zFetchURL($newurl, $binary, $redirects, $opts);
 237                         }
 238                 }
 239
 240                 $a->set_curl_code($http_code);
 241                 $a->set_curl_content_type($curl_info['content_type']);
 242
 243                 $rc = intval($http_code);
 244                 $ret['return_code'] = $rc;
 245                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 246                 $ret['redirect_url'] = $url;
 247
 248                 if (!$ret['success']) {
 249                         $ret['error'] = curl_error($ch);
 250                         $ret['debug'] = $curl_info;
 251                         logger('z_fetch_url: error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 252                         logger('z_fetch_url: debug: '.print_r($curl_info, true), LOGGER_DATA);
 253                 }
 254
 255                 $ret['body'] = substr($s, strlen($header));
 256                 $ret['header'] = $header;
 257
 258                 if (x($opts, 'debug')) {
 259                         $ret['debug'] = $curl_info;
 260                 }
 261
 262                 @curl_close($ch);
 263
 264                 $a->save_timestamp($stamp1, 'network');
 265
 266                 return($ret);
 267         }
 268
 269         /**
 270          * @brief Send POST request to $url
 271          *
 272          * @param string  $url       URL to post
 273          * @param mixed   $params    array of POST variables
 274          * @param string  $headers   HTTP headers
 275          * @param integer $redirects Recursion counter for internal use - default = 0
 276          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 277          *
 278          * @return string The content
 279          */
 280         public static function postURL($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 281         {
 282                 $stamp1 = microtime(true);
 283
 284                 if (blocked_url($url)) {
 285                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 286                         return false;
 287                 }
 288
 289                 $a = get_app();
 290                 $ch = curl_init($url);
 291
 292                 if (($redirects > 8) || (!$ch)) {
 293                         return false;
 294                 }
 295
 296                 logger('post_url: start ' . $url, LOGGER_DATA);
 297
 298                 curl_setopt($ch, CURLOPT_HEADER, true);
 299                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 300                 curl_setopt($ch, CURLOPT_POST, 1);
 301                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 302                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 303
 304                 if (Config::get('system', 'ipv4_resolve', false)) {
 305                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 306                 }
 307
 308                 if (intval($timeout)) {
 309                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 310                 } else {
 311                         $curl_time = Config::get('system', 'curl_timeout', 60);
 312                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 313                 }
 314
 315                 if (defined('LIGHTTPD')) {
 316                         if (!is_array($headers)) {
 317                                 $headers = ['Expect:'];
 318                         } else {
 319                                 if (!in_array('Expect:', $headers)) {
 320                                         array_push($headers, 'Expect:');
 321                                 }
 322                         }
 323                 }
 324
 325                 if ($headers) {
 326                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 327                 }
 328
 329                 $check_cert = Config::get('system', 'verifyssl');
 330                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 331
 332                 if ($check_cert) {
 333                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 334                 }
 335
 336                 $proxy = Config::get('system', 'proxy');
 337
 338                 if (strlen($proxy)) {
 339                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 340                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 341                         $proxyuser = Config::get('system', 'proxyuser');
 342                         if (strlen($proxyuser)) {
 343                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 344                         }
 345                 }
 346
 347                 $a->set_curl_code(0);
 348
 349                 // don't let curl abort the entire application
 350                 // if it throws any errors.
 351
 352                 $s = @curl_exec($ch);
 353
 354                 $base = $s;
 355                 $curl_info = curl_getinfo($ch);
 356                 $http_code = $curl_info['http_code'];
 357
 358                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 359
 360                 $header = '';
 361
 362                 // Pull out multiple headers, e.g. proxy and continuation headers
 363                 // allow for HTTP/2.x without fixing code
 364
 365                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 366                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 367                         $header .= $chunk;
 368                         $base = substr($base, strlen($chunk));
 369                 }
 370
 371                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 372                         $matches = [];
 373                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 374                         $newurl = trim(array_pop($matches));
 375
 376                         if (strpos($newurl, '/') === 0) {
 377                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 378                         }
 379
 380                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 381                                 $redirects++;
 382                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 383                                 return self::postURL($newurl, $params, $headers, $redirects, $timeout);
 384                         }
 385                 }
 386
 387                 $a->set_curl_code($http_code);
 388
 389                 $body = substr($s, strlen($header));
 390
 391                 $a->set_curl_headers($header);
 392
 393                 curl_close($ch);
 394
 395                 $a->save_timestamp($stamp1, 'network');
 396
 397                 logger('post_url: end ' . $url, LOGGER_DATA);
 398
 399                 return $body;
 400         }
 401
 402         /**
 403          * Generic XML return
 404          * Outputs a basic dfrn XML status structure to STDOUT, with a <status> variable
 405          * of $st and an optional text <message> of $message and terminates the current process.
 406          */
 407         public static function xmlStatus($st, $message = '')
 408         {
 409                 $result = ['status' => $st];
 410
 411                 if ($message != '') {
 412                         $result['message'] = $message;
 413                 }
 414
 415                 if ($st) {
 416                         logger('xml_status returning non_zero: ' . $st . " message=" . $message);
 417                 }
 418
 419                 header("Content-type: text/xml");
 420
 421                 $xmldata = ["result" => $result];
 422
 423                 echo XML::fromArray($xmldata, $xml);
 424
 425                 killme();
 426         }
 427
 428         /**
 429          * @brief Send HTTP status header and exit.
 430          *
 431          * @param integer $val         HTTP status result value
 432          * @param array   $description optional message
 433          *                             'title' => header title
 434          *                             'description' => optional message
 435          */
 436         public static function httpStatusExit($val, $description = [])
 437         {
 438                 $err = '';
 439                 if ($val >= 400) {
 440                         $err = 'Error';
 441                         if (!isset($description["title"])) {
 442                                 $description["title"] = $err." ".$val;
 443                         }
 444                 }
 445
 446                 if ($val >= 200 && $val < 300) {
 447                         $err = 'OK';
 448                 }
 449
 450                 logger('http_status_exit ' . $val);
 451                 header($_SERVER["SERVER_PROTOCOL"] . ' ' . $val . ' ' . $err);
 452
 453                 if (isset($description["title"])) {
 454                         $tpl = get_markup_template('http_status.tpl');
 455                         echo replace_macros(
 456                                 $tpl,
 457                                 [
 458                                         '$title' => $description["title"],
 459                                         '$description' => $description["description"]]
 460                         );
 461                 }
 462
 463                 killme();
 464         }
 465
 466         /**
 467          * @brief Check URL to se if ts's real
 468          *
 469          * Take a URL from the wild, prepend http:// if necessary
 470          * and check DNS to see if it's real (or check if is a valid IP address)
 471          *
 472          * @param string $url The URL to be validated
 473          * @return string|boolean The actual working URL, false else
 474          */
 475         public static function validateURL($url)
 476         {
 477                 if (Config::get('system', 'disable_url_validation')) {
 478                         return $url;
 479                 }
 480
 481                 // no naked subdomains (allow localhost for tests)
 482                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 483                         return false;
 484                 }
 485
 486                 if (substr($url, 0, 4) != 'http') {
 487                         $url = 'http://' . $url;
 488                 }
 489
 490                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 491                 $h = @parse_url($url);
 492
 493                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 494                         return $url;
 495                 }
 496
 497                 return false;
 498         }
 499
 500         /**
 501          * @brief Checks that email is an actual resolvable internet address
 502          *
 503          * @param string $addr The email address
 504          * @return boolean True if it's a valid email address, false if it's not
 505          */
 506         public static function validateEmail($addr)
 507         {
 508                 if (Config::get('system', 'disable_email_validation')) {
 509                         return true;
 510                 }
 511
 512                 if (! strpos($addr, '@')) {
 513                         return false;
 514                 }
 515
 516                 $h = substr($addr, strpos($addr, '@') + 1);
 517
 518                 if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 519                         return true;
 520                 }
 521                 return false;
 522         }
 523
 524         /**
 525          * @brief Check if URL is allowed
 526          *
 527          * Check $url against our list of allowed sites,
 528          * wildcards allowed. If allowed_sites is unset return true;
 529          *
 530          * @param string $url URL which get tested
 531          * @return boolean True if url is allowed otherwise return false
 532          */
 533         public static function allowedURL($url)
 534         {
 535                 $h = @parse_url($url);
 536
 537                 if (! $h) {
 538                         return false;
 539                 }
 540
 541                 $str_allowed = Config::get('system', 'allowed_sites');
 542                 if (! $str_allowed) {
 543                         return true;
 544                 }
 545
 546                 $found = false;
 547
 548                 $host = strtolower($h['host']);
 549
 550                 // always allow our own site
 551                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 552                         return true;
 553                 }
 554
 555                 $fnmatch = function_exists('fnmatch');
 556                 $allowed = explode(',', $str_allowed);
 557
 558                 if (count($allowed)) {
 559                         foreach ($allowed as $a) {
 560                                 $pat = strtolower(trim($a));
 561                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 562                                         $found = true;
 563                                         break;
 564                                 }
 565                         }
 566                 }
 567                 return $found;
 568         }
 569
 570         /**
 571          * Checks if the provided url domain is on the domain blocklist.
 572          * Returns true if it is or malformed URL, false if not.
 573          *
 574          * @param string $url The url to check the domain from
 575          *
 576          * @return boolean
 577          */
 578         public static function blockedURL($url)
 579         {
 580                 $h = @parse_url($url);
 581
 582                 if (! $h) {
 583                         return true;
 584                 }
 585
 586                 $domain_blocklist = Config::get('system', 'blocklist', []);
 587                 if (! $domain_blocklist) {
 588                         return false;
 589                 }
 590
 591                 $host = strtolower($h['host']);
 592
 593                 foreach ($domain_blocklist as $domain_block) {
 594                         if (strtolower($domain_block['domain']) == $host) {
 595                                 return true;
 596                         }
 597                 }
 598
 599                 return false;
 600         }
 601
 602         /**
 603          * @brief Check if email address is allowed to register here.
 604          *
 605          * Compare against our list (wildcards allowed).
 606          *
 607          * @param  string $email email address
 608          * @return boolean False if not allowed, true if allowed
 609          *    or if allowed list is not configured
 610          */
 611         public static function allowedEmail($email)
 612         {
 613                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 614                 if (!$domain) {
 615                         return false;
 616                 }
 617
 618                 $str_allowed = Config::get('system', 'allowed_email', '');
 619                 if (!x($str_allowed)) {
 620                         return true;
 621                 }
 622
 623                 $allowed = explode(',', $str_allowed);
 624
 625                 return allowed_domain($domain, $allowed);
 626         }
 627
 628         /**
 629          * Checks for the existence of a domain in a domain list
 630          *
 631          * @brief Checks for the existence of a domain in a domain list
 632          * @param string $domain
 633          * @param array  $domain_list
 634          * @return boolean
 635          */
 636         public static function allowedDomain($domain, array $domain_list)
 637         {
 638                 $found = false;
 639
 640                 foreach ($domain_list as $item) {
 641                         $pat = strtolower(trim($item));
 642                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 643                                 $found = true;
 644                                 break;
 645                         }
 646                 }
 647
 648                 return $found;
 649         }
 650
 651         public static function avatarImg($email)
 652         {
 653                 $avatar['size'] = 175;
 654                 $avatar['email'] = $email;
 655                 $avatar['url'] = '';
 656                 $avatar['success'] = false;
 657
 658                 Addon::callHooks('avatar_lookup', $avatar);
 659
 660                 if (! $avatar['success']) {
 661                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 662                 }
 663
 664                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 665                 return $avatar['url'];
 666         }
 667
 668         public static function parseXmlString($s, $strict = true)
 669         {
 670                 // the "strict" parameter is deactivated
 671
 672                 /// @todo Move this function to the xml class
 673                 libxml_use_internal_errors(true);
 674
 675                 $x = @simplexml_load_string($s);
 676                 if (!$x) {
 677                         logger('libxml: parse: error: ' . $s, LOGGER_DATA);
 678                         foreach (libxml_get_errors() as $err) {
 679                                 logger('libxml: parse: ' . $err->code." at ".$err->line.":".$err->column." : ".$err->message, LOGGER_DATA);
 680                         }
 681                         libxml_clear_errors();
 682                 }
 683                 return $x;
 684         }
 685
 686         public static function scaleExternalImages($srctext, $include_link = true, $scale_replace = false)
 687         {
 688                 // Suppress "view full size"
 689                 if (intval(Config::get('system', 'no_view_full_size'))) {
 690                         $include_link = false;
 691                 }
 692
 693                 // Picture addresses can contain special characters
 694                 $s = htmlspecialchars_decode($srctext);
 695
 696                 $matches = null;
 697                 $c = preg_match_all('/\[img.*?\](.*?)\[\/img\]/ism', $s, $matches, PREG_SET_ORDER);
 698                 if ($c) {
 699                         foreach ($matches as $mtch) {
 700                                 logger('scale_external_image: ' . $mtch[1]);
 701
 702                                 $hostname = str_replace('www.', '', substr(System::baseUrl(), strpos(System::baseUrl(), '://') + 3));
 703                                 if (stristr($mtch[1], $hostname)) {
 704                                         continue;
 705                                 }
 706
 707                                 // $scale_replace, if passed, is an array of two elements. The
 708                                 // first is the name of the full-size image. The second is the
 709                                 // name of a remote, scaled-down version of the full size image.
 710                                 // This allows Friendica to display the smaller remote image if
 711                                 // one exists, while still linking to the full-size image
 712                                 if ($scale_replace) {
 713                                         $scaled = str_replace($scale_replace[0], $scale_replace[1], $mtch[1]);
 714                                 } else {
 715                                         $scaled = $mtch[1];
 716                                 }
 717                                 $i = self::fetchURL($scaled);
 718                                 if (! $i) {
 719                                         return $srctext;
 720                                 }
 721
 722                                 // guess mimetype from headers or filename
 723                                 $type = Image::guessType($mtch[1], true);
 724
 725                                 if ($i) {
 726                                         $Image = new Image($i, $type);
 727                                         if ($Image->isValid()) {
 728                                                 $orig_width = $Image->getWidth();
 729                                                 $orig_height = $Image->getHeight();
 730
 731                                                 if ($orig_width > 640 || $orig_height > 640) {
 732                                                         $Image->scaleDown(640);
 733                                                         $new_width = $Image->getWidth();
 734                                                         $new_height = $Image->getHeight();
 735                                                         logger('scale_external_images: ' . $orig_width . '->' . $new_width . 'w ' . $orig_height . '->' . $new_height . 'h' . ' match: ' . $mtch[0], LOGGER_DEBUG);
 736                                                         $s = str_replace(
 737                                                                 $mtch[0],
 738                                                                 '[img=' . $new_width . 'x' . $new_height. ']' . $scaled . '[/img]'
 739                                                                 . "\n" . (($include_link)
 740                                                                         ? '[url=' . $mtch[1] . ']' . L10n::t('view full size') . '[/url]' . "\n"
 741                                                                         : ''),
 742                                                                 $s
 743                                                         );
 744                                                         logger('scale_external_images: new string: ' . $s, LOGGER_DEBUG);
 745                                                 }
 746                                         }
 747                                 }
 748                         }
 749                 }
 750
 751                 // replace the special char encoding
 752                 $s = htmlspecialchars($s, ENT_NOQUOTES, 'UTF-8');
 753                 return $s;
 754         }
 755
 756         public static function fixContactSslPolicy(&$contact, $new_policy)
 757         {
 758                 $ssl_changed = false;
 759                 if ((intval($new_policy) == SSL_POLICY_SELFSIGN || $new_policy === 'self') && strstr($contact['url'], 'https:')) {
 760                         $ssl_changed = true;
 761                         $contact['url']     =   str_replace('https:', 'http:', $contact['url']);
 762                         $contact['request'] =   str_replace('https:', 'http:', $contact['request']);
 763                         $contact['notify']  =   str_replace('https:', 'http:', $contact['notify']);
 764                         $contact['poll']    =   str_replace('https:', 'http:', $contact['poll']);
 765                         $contact['confirm'] =   str_replace('https:', 'http:', $contact['confirm']);
 766                         $contact['poco']    =   str_replace('https:', 'http:', $contact['poco']);
 767                 }
 768
 769                 if ((intval($new_policy) == SSL_POLICY_FULL || $new_policy === 'full') && strstr($contact['url'], 'http:')) {
 770                         $ssl_changed = true;
 771                         $contact['url']     =   str_replace('http:', 'https:', $contact['url']);
 772                         $contact['request'] =   str_replace('http:', 'https:', $contact['request']);
 773                         $contact['notify']  =   str_replace('http:', 'https:', $contact['notify']);
 774                         $contact['poll']    =   str_replace('http:', 'https:', $contact['poll']);
 775                         $contact['confirm'] =   str_replace('http:', 'https:', $contact['confirm']);
 776                         $contact['poco']    =   str_replace('http:', 'https:', $contact['poco']);
 777                 }
 778
 779                 if ($ssl_changed) {
 780                         $fields = ['url' => $contact['url'], 'request' => $contact['request'],
 781                                         'notify' => $contact['notify'], 'poll' => $contact['poll'],
 782                                         'confirm' => $contact['confirm'], 'poco' => $contact['poco']];
 783                         dba::update('contact', $fields, ['id' => $contact['id']]);
 784                 }
 785         }
 786
 787         /**
 788          * @brief Remove Google Analytics and other tracking platforms params from URL
 789          *
 790          * @param string $url Any user-submitted URL that may contain tracking params
 791          * @return string The same URL stripped of tracking parameters
 792          */
 793         public static function stripTrackingQueryParams($url)
 794         {
 795                 $urldata = parse_url($url);
 796                 if (is_string($urldata["query"])) {
 797                         $query = $urldata["query"];
 798                         parse_str($query, $querydata);
 799
 800                         if (is_array($querydata)) {
 801                                 foreach ($querydata as $param => $value) {
 802                                         if (in_array(
 803                                                 $param,
 804                                                 [
 805                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 806                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 807                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 808                                                         "awesm", "wtrid",
 809                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 810                                                 )
 811                                         ) {
 812                                                 $pair = $param . "=" . urlencode($value);
 813                                                 $url = str_replace($pair, "", $url);
 814
 815                                                 // Second try: if the url isn't encoded completely
 816                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 817                                                 $url = str_replace($pair, "", $url);
 818
 819                                                 // Third try: Maybey the url isn't encoded at all
 820                                                 $pair = $param . "=" . $value;
 821                                                 $url = str_replace($pair, "", $url);
 822
 823                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 824                                         }
 825                                 }
 826                         }
 827
 828                         if (substr($url, -1, 1) == "?") {
 829                                 $url = substr($url, 0, -1);
 830                         }
 831                 }
 832
 833                 return $url;
 834         }
 835
 836         /**
 837          * @brief Returns the original URL of the provided URL
 838          *
 839          * This function strips tracking query params and follows redirections, either
 840          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 841          *
 842          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 843          *
 844          * @see ParseUrl::getSiteinfo
 845          *
 846          * @param string $url       A user-submitted URL
 847          * @param int    $depth     The current redirection recursion level (internal)
 848          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 849          * @return string A canonical URL
 850          */
 851         public static function originalURL($url, $depth = 1, $fetchbody = false)
 852         {
 853                 $a = get_app();
 854
 855                 $url = strip_tracking_query_params($url);
 856
 857                 if ($depth > 10) {
 858                         return($url);
 859                 }
 860
 861                 $url = trim($url, "'");
 862
 863                 $stamp1 = microtime(true);
 864
 865                 $ch = curl_init();
 866                 curl_setopt($ch, CURLOPT_URL, $url);
 867                 curl_setopt($ch, CURLOPT_HEADER, 1);
 868                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 869                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 870                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 871                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 872
 873                 curl_exec($ch);
 874                 $curl_info = @curl_getinfo($ch);
 875                 $http_code = $curl_info['http_code'];
 876                 curl_close($ch);
 877
 878                 $a->save_timestamp($stamp1, "network");
 879
 880                 if ($http_code == 0) {
 881                         return($url);
 882                 }
 883
 884                 if ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 885                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 886                 ) {
 887                         if ($curl_info['redirect_url'] != "") {
 888                                 return(original_url($curl_info['redirect_url'], ++$depth, $fetchbody));
 889                         } else {
 890                                 return(original_url($curl_info['location'], ++$depth, $fetchbody));
 891                         }
 892                 }
 893
 894                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 895                 if (!$fetchbody) {
 896                         return(original_url($url, ++$depth, true));
 897                 }
 898
 899                 // if the file is too large then exit
 900                 if ($curl_info["download_content_length"] > 1000000) {
 901                         return($url);
 902                 }
 903
 904                 // if it isn't a HTML file then exit
 905                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 906                         return($url);
 907                 }
 908
 909                 $stamp1 = microtime(true);
 910
 911                 $ch = curl_init();
 912                 curl_setopt($ch, CURLOPT_URL, $url);
 913                 curl_setopt($ch, CURLOPT_HEADER, 0);
 914                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 915                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 916                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 917                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 918
 919                 $body = curl_exec($ch);
 920                 curl_close($ch);
 921
 922                 $a->save_timestamp($stamp1, "network");
 923
 924                 if (trim($body) == "") {
 925                         return($url);
 926                 }
 927
 928                 // Check for redirect in meta elements
 929                 $doc = new DOMDocument();
 930                 @$doc->loadHTML($body);
 931
 932                 $xpath = new DomXPath($doc);
 933
 934                 $list = $xpath->query("//meta[@content]");
 935                 foreach ($list as $node) {
 936                         $attr = [];
 937                         if ($node->attributes->length) {
 938                                 foreach ($node->attributes as $attribute) {
 939                                         $attr[$attribute->name] = $attribute->value;
 940                                 }
 941                         }
 942
 943                         if (@$attr["http-equiv"] == 'refresh') {
 944                                 $path = $attr["content"];
 945                                 $pathinfo = explode(";", $path);
 946                                 foreach ($pathinfo as $value) {
 947                                         if (substr(strtolower($value), 0, 4) == "url=") {
 948                                                 return(original_url(substr($value, 4), ++$depth));
 949                                         }
 950                                 }
 951                         }
 952                 }
 953
 954                 return $url;
 955         }
 956
 957         public static function shortLink($url)
 958         {
 959                 $slinky = new Slinky($url);
 960                 $yourls_url = Config::get('yourls', 'url1');
 961                 if ($yourls_url) {
 962                         $yourls_username = Config::get('yourls', 'username1');
 963                         $yourls_password = Config::get('yourls', 'password1');
 964                         $yourls_ssl = Config::get('yourls', 'ssl1');
 965                         $yourls = new Slinky_YourLS();
 966                         $yourls->set('username', $yourls_username);
 967                         $yourls->set('password', $yourls_password);
 968                         $yourls->set('ssl', $yourls_ssl);
 969                         $yourls->set('yourls-url', $yourls_url);
 970                         $slinky->set_cascade([$yourls, new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 971                 } else {
 972                         // setup a cascade of shortening services
 973                         // try to get a short link from these services
 974                         // in the order ur1.ca, tinyurl
 975                         $slinky->set_cascade([new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 976                 }
 977                 return $slinky->short();
 978         }
 979
 980         /**
 981          * @brief Encodes content to json
 982          *
 983          * This function encodes an array to json format
 984          * and adds an application/json HTTP header to the output.
 985          * After finishing the process is getting killed.
 986          *
 987          * @param array $x The input content
 988          */
 989         public static function jsonReturnAndDie($x)
 990         {
 991                 header("content-type: application/json");
 992                 echo json_encode($x);
 993                 killme();
 994         }
 995
 996         /**
 997          * @brief Find the matching part between two url
 998          *
 999          * @param string $url1
1000          * @param string $url2
1001          * @return string The matching part
1002          */
1003         public static function matchingURL($url1, $url2)
1004         {
1005                 if (($url1 == "") || ($url2 == "")) {
1006                         return "";
1007                 }
1008
1009                 $url1 = normalise_link($url1);
1010                 $url2 = normalise_link($url2);
1011
1012                 $parts1 = parse_url($url1);
1013                 $parts2 = parse_url($url2);
1014
1015                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
1016                         return "";
1017                 }
1018
1019                 if ($parts1["scheme"] != $parts2["scheme"]) {
1020                         return "";
1021                 }
1022
1023                 if ($parts1["host"] != $parts2["host"]) {
1024                         return "";
1025                 }
1026
1027                 if ($parts1["port"] != $parts2["port"]) {
1028                         return "";
1029                 }
1030
1031                 $match = $parts1["scheme"]."://".$parts1["host"];
1032
1033                 if ($parts1["port"]) {
1034                         $match .= ":".$parts1["port"];
1035                 }
1036
1037                 $pathparts1 = explode("/", $parts1["path"]);
1038                 $pathparts2 = explode("/", $parts2["path"]);
1039
1040                 $i = 0;
1041                 $path = "";
1042                 do {
1043                         $path1 = $pathparts1[$i];
1044                         $path2 = $pathparts2[$i];
1045
1046                         if ($path1 == $path2) {
1047                                 $path .= $path1."/";
1048                         }
1049                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
1050
1051                 $match .= $path;
1052
1053                 return normalise_link($match);
1054         }
1055
1056         /**
1057          * @brief Glue url parts together
1058          *
1059          * @param array $parsed URL parts
1060          *
1061          * @return string The glued URL
1062          */
1063         public static function unParseURL($parsed)
1064         {
1065                 $get = function ($key) use ($parsed) {
1066                         return isset($parsed[$key]) ? $parsed[$key] : null;
1067                 };
1068
1069                 $pass      = $get('pass');
1070                 $user      = $get('user');
1071                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
1072                 $port      = $get('port');
1073                 $scheme    = $get('scheme');
1074                 $query     = $get('query');
1075                 $fragment  = $get('fragment');
1076                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
1077                                                 $get('host') .
1078                                                 ($port ? ":$port" : '');
1079
1080                 return  (strlen($scheme) ? $scheme.":" : '') .
1081                         (strlen($authority) ? "//".$authority : '') .
1082                         $get('path') .
1083                         (strlen($query) ? "?".$query : '') .
1084                         (strlen($fragment) ? "#".$fragment : '');
1085         }
1086 }