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 use DOMDocument;
  16 use DomXPath;
  17
  18 class Network
  19 {
  20         /**
  21          * 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          * @brief Curl wrapper
  28          * @param string  $url            URL to fetch
  29          * @param boolean $binary         default false
  30          *                                TRUE if asked to return binary results (file download)
  31          * @param integer $redirects      The recursion counter for internal use - default 0
  32          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  33          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  34          * @param string  $cookiejar      Path to cookie jar file
  35          *
  36          * @return string The fetched content
  37          */
  38         public static function fetchUrl($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  39         {
  40                 $ret = self::fetchUrlFull($url, $binary, $redirects, $timeout, $accept_content, $cookiejar);
  41
  42                 return $ret['body'];
  43         }
  44
  45         /**
  46          * Curl wrapper with array of return values.
  47          *
  48          * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
  49          * all the information collected during the fetch.
  50          *
  51          * @brief Curl wrapper with array of return values.
  52          * @param string  $url            URL to fetch
  53          * @param boolean $binary         default false
  54          *                                TRUE if asked to return binary results (file download)
  55          * @param integer $redirects      The recursion counter for internal use - default 0
  56          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  57          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  58          * @param string  $cookiejar      Path to cookie jar file
  59          *
  60          * @return array With all relevant information, 'body' contains the actual fetched content.
  61          */
  62         public static function fetchUrlFull($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  63         {
  64                 return self::curl(
  65                         $url,
  66                         $binary,
  67                         $redirects,
  68                         ['timeout'=>$timeout,
  69                         'accept_content'=>$accept_content,
  70                         'cookiejar'=>$cookiejar
  71                         ]
  72                 );
  73         }
  74
  75         /**
  76          * @brief fetches an URL.
  77          *
  78          * @param string  $url       URL to fetch
  79          * @param boolean $binary    default false
  80          *                           TRUE if asked to return binary results (file download)
  81          * @param int     $redirects The recursion counter for internal use - default 0
  82          * @param array   $opts      (optional parameters) assoziative array with:
  83          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  84          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  85          *                           'http_auth' => username:password
  86          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  87          *                           'nobody' => only return the header
  88          *                           'cookiejar' => path to cookie jar file
  89          *
  90          * @return array an assoziative array with:
  91          *    int 'return_code' => HTTP return code or 0 if timeout or failure
  92          *    boolean 'success' => boolean true (if HTTP 2xx result) or false
  93          *    string 'redirect_url' => in case of redirect, content was finally retrieved from this URL
  94          *    string 'header' => HTTP headers
  95          *    string 'body' => fetched content
  96          */
  97         public static function curl($url, $binary = false, &$redirects = 0, $opts = [])
  98         {
  99                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
 100
 101                 $stamp1 = microtime(true);
 102
 103                 $a = get_app();
 104
 105                 $parts = parse_url($url);
 106                 $path_parts = explode('/', defaults($parts, 'path', ''));
 107                 foreach ($path_parts as $part) {
 108                         if (strlen($part) <> mb_strlen($part)) {
 109                                 $parts2[] = rawurlencode($part);
 110                         } else {
 111                                 $parts2[] = $part;
 112                         }
 113                 }
 114                 $parts['path'] =  implode('/', $parts2);
 115                 $url = self::unparseURL($parts);
 116
 117                 if (self::isUrlBlocked($url)) {
 118                         logger('domain of ' . $url . ' is blocked', LOGGER_DATA);
 119                         return $ret;
 120                 }
 121
 122                 $ch = @curl_init($url);
 123
 124                 if (($redirects > 8) || (!$ch)) {
 125                         return $ret;
 126                 }
 127
 128                 @curl_setopt($ch, CURLOPT_HEADER, true);
 129
 130                 if (x($opts, "cookiejar")) {
 131                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 132                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 133                 }
 134
 135                 // These settings aren't needed. We're following the location already.
 136                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 137                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 138
 139                 if (x($opts, 'accept_content')) {
 140                         curl_setopt(
 141                                 $ch,
 142                                 CURLOPT_HTTPHEADER,
 143                                 ['Accept: ' . $opts['accept_content']]
 144                         );
 145                 }
 146
 147                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 148                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 149
 150                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 151
 152                 if ($range > 0) {
 153                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 154                 }
 155
 156                 // Without this setting it seems as if some webservers send compressed content
 157                 // This seems to confuse curl so that it shows this uncompressed.
 158                 /// @todo  We could possibly set this value to "gzip" or something similar
 159                 curl_setopt($ch, CURLOPT_ENCODING, '');
 160
 161                 if (x($opts, 'headers')) {
 162                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 163                 }
 164
 165                 if (x($opts, 'nobody')) {
 166                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 167                 }
 168
 169                 if (x($opts, 'timeout')) {
 170                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 171                 } else {
 172                         $curl_time = Config::get('system', 'curl_timeout', 60);
 173                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 174                 }
 175
 176                 // by default we will allow self-signed certs
 177                 // but you can override this
 178
 179                 $check_cert = Config::get('system', 'verifyssl');
 180                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 181
 182                 if ($check_cert) {
 183                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 184                 }
 185
 186                 $proxy = Config::get('system', 'proxy');
 187
 188                 if (strlen($proxy)) {
 189                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 190                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 191                         $proxyuser = @Config::get('system', 'proxyuser');
 192
 193                         if (strlen($proxyuser)) {
 194                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 195                         }
 196                 }
 197
 198                 if (Config::get('system', 'ipv4_resolve', false)) {
 199                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 200                 }
 201
 202                 if ($binary) {
 203                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 204                 }
 205
 206                 $a->set_curl_code(0);
 207
 208                 // don't let curl abort the entire application
 209                 // if it throws any errors.
 210
 211                 $s = @curl_exec($ch);
 212                 $curl_info = @curl_getinfo($ch);
 213
 214                 // Special treatment for HTTP Code 416
 215                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 216                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 217                         @curl_setopt($ch, CURLOPT_RANGE, '');
 218                         $s = @curl_exec($ch);
 219                         $curl_info = @curl_getinfo($ch);
 220                 }
 221
 222                 if (curl_errno($ch) !== CURLE_OK) {
 223                         logger('error fetching ' . $url . ': ' . curl_error($ch), LOGGER_INFO);
 224                 }
 225
 226                 $ret['errno'] = curl_errno($ch);
 227
 228                 $base = $s;
 229                 $ret['info'] = $curl_info;
 230
 231                 $http_code = $curl_info['http_code'];
 232
 233                 logger($url . ': ' . $http_code . " " . $s, LOGGER_DATA);
 234                 $header = '';
 235
 236                 // Pull out multiple headers, e.g. proxy and continuation headers
 237                 // allow for HTTP/2.x without fixing code
 238
 239                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 240                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 241                         $header .= $chunk;
 242                         $base = substr($base, strlen($chunk));
 243                 }
 244
 245                 $a->set_curl_code($http_code);
 246                 $a->set_curl_content_type($curl_info['content_type']);
 247                 $a->set_curl_headers($header);
 248
 249                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 250                         $new_location_info = @parse_url($curl_info['redirect_url']);
 251                         $old_location_info = @parse_url($curl_info['url']);
 252
 253                         $newurl = $curl_info['redirect_url'];
 254
 255                         if (empty($new_location_info['path']) && !empty($new_location_info['host'])) {
 256                                 $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
 257                         }
 258
 259                         $matches = [];
 260
 261                         if (preg_match('/(Location:|URI:)(.*?)\n/i', $header, $matches)) {
 262                                 $newurl = trim(array_pop($matches));
 263                         }
 264                         if (strpos($newurl, '/') === 0) {
 265                                 $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
 266                         }
 267                         $old_location_query = @parse_url($url, PHP_URL_QUERY);
 268
 269                         if ($old_location_query != '') {
 270                                 $newurl .= '?' . $old_location_query;
 271                         }
 272
 273                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 274                                 $redirects++;
 275                                 @curl_close($ch);
 276                                 return self::curl($newurl, $binary, $redirects, $opts);
 277                         }
 278                 }
 279
 280                 $a->set_curl_code($http_code);
 281                 $a->set_curl_content_type($curl_info['content_type']);
 282
 283                 $rc = intval($http_code);
 284                 $ret['return_code'] = $rc;
 285                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 286                 $ret['redirect_url'] = $url;
 287
 288                 if (!$ret['success']) {
 289                         $ret['error'] = curl_error($ch);
 290                         $ret['debug'] = $curl_info;
 291                         logger('error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 292                         logger('debug: '.print_r($curl_info, true), LOGGER_DATA);
 293                 }
 294
 295                 $ret['body'] = substr($s, strlen($header));
 296                 $ret['header'] = $header;
 297
 298                 if (x($opts, 'debug')) {
 299                         $ret['debug'] = $curl_info;
 300                 }
 301
 302                 @curl_close($ch);
 303
 304                 $a->save_timestamp($stamp1, 'network');
 305
 306                 return($ret);
 307         }
 308
 309         /**
 310          * @brief Send POST request to $url
 311          *
 312          * @param string  $url       URL to post
 313          * @param mixed   $params    array of POST variables
 314          * @param string  $headers   HTTP headers
 315          * @param integer $redirects Recursion counter for internal use - default = 0
 316          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 317          *
 318          * @return string The content
 319          */
 320         public static function post($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 321         {
 322                 $stamp1 = microtime(true);
 323
 324                 if (self::isUrlBlocked($url)) {
 325                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 326                         return false;
 327                 }
 328
 329                 $a = get_app();
 330                 $ch = curl_init($url);
 331
 332                 if (($redirects > 8) || (!$ch)) {
 333                         return false;
 334                 }
 335
 336                 logger('post_url: start ' . $url, LOGGER_DATA);
 337
 338                 curl_setopt($ch, CURLOPT_HEADER, true);
 339                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 340                 curl_setopt($ch, CURLOPT_POST, 1);
 341                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 342                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 343
 344                 if (Config::get('system', 'ipv4_resolve', false)) {
 345                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 346                 }
 347
 348                 if (intval($timeout)) {
 349                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 350                 } else {
 351                         $curl_time = Config::get('system', 'curl_timeout', 60);
 352                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 353                 }
 354
 355                 if (defined('LIGHTTPD')) {
 356                         if (!is_array($headers)) {
 357                                 $headers = ['Expect:'];
 358                         } else {
 359                                 if (!in_array('Expect:', $headers)) {
 360                                         array_push($headers, 'Expect:');
 361                                 }
 362                         }
 363                 }
 364
 365                 if ($headers) {
 366                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 367                 }
 368
 369                 $check_cert = Config::get('system', 'verifyssl');
 370                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 371
 372                 if ($check_cert) {
 373                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 374                 }
 375
 376                 $proxy = Config::get('system', 'proxy');
 377
 378                 if (strlen($proxy)) {
 379                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 380                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 381                         $proxyuser = Config::get('system', 'proxyuser');
 382                         if (strlen($proxyuser)) {
 383                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 384                         }
 385                 }
 386
 387                 $a->set_curl_code(0);
 388
 389                 // don't let curl abort the entire application
 390                 // if it throws any errors.
 391
 392                 $s = @curl_exec($ch);
 393
 394                 $base = $s;
 395                 $curl_info = curl_getinfo($ch);
 396                 $http_code = $curl_info['http_code'];
 397
 398                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 399
 400                 $header = '';
 401
 402                 // Pull out multiple headers, e.g. proxy and continuation headers
 403                 // allow for HTTP/2.x without fixing code
 404
 405                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 406                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 407                         $header .= $chunk;
 408                         $base = substr($base, strlen($chunk));
 409                 }
 410
 411                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 412                         $matches = [];
 413                         $new_location_info = @parse_url($curl_info['redirect_url']);
 414                         $old_location_info = @parse_url($curl_info['url']);
 415
 416                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 417                         $newurl = trim(array_pop($matches));
 418
 419                         if (strpos($newurl, '/') === 0) {
 420                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 421                         }
 422
 423                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 424                                 $redirects++;
 425                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 426                                 return self::post($newurl, $params, $headers, $redirects, $timeout);
 427                         }
 428                 }
 429
 430                 $a->set_curl_code($http_code);
 431
 432                 $body = substr($s, strlen($header));
 433
 434                 $a->set_curl_headers($header);
 435
 436                 curl_close($ch);
 437
 438                 $a->save_timestamp($stamp1, 'network');
 439
 440                 logger('post_url: end ' . $url, LOGGER_DATA);
 441
 442                 return $body;
 443         }
 444
 445         /**
 446          * @brief Check URL to see if it's real
 447          *
 448          * Take a URL from the wild, prepend http:// if necessary
 449          * and check DNS to see if it's real (or check if is a valid IP address)
 450          *
 451          * @param string $url The URL to be validated
 452          * @return string|boolean The actual working URL, false else
 453          */
 454         public static function isUrlValid($url)
 455         {
 456                 if (Config::get('system', 'disable_url_validation')) {
 457                         return $url;
 458                 }
 459
 460                 // no naked subdomains (allow localhost for tests)
 461                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 462                         return false;
 463                 }
 464
 465                 if (substr($url, 0, 4) != 'http') {
 466                         $url = 'http://' . $url;
 467                 }
 468
 469                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 470                 $h = @parse_url($url);
 471
 472                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 473                         return $url;
 474                 }
 475
 476                 return false;
 477         }
 478
 479         /**
 480          * @brief Checks that email is an actual resolvable internet address
 481          *
 482          * @param string $addr The email address
 483          * @return boolean True if it's a valid email address, false if it's not
 484          */
 485         public static function isEmailDomainValid($addr)
 486         {
 487                 if (Config::get('system', 'disable_email_validation')) {
 488                         return true;
 489                 }
 490
 491                 if (! strpos($addr, '@')) {
 492                         return false;
 493                 }
 494
 495                 $h = substr($addr, strpos($addr, '@') + 1);
 496
 497                 // Concerning the @ see here: https://stackoverflow.com/questions/36280957/dns-get-record-a-temporary-server-error-occurred
 498                 if ($h && (@dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 499                         return true;
 500                 }
 501                 if ($h && @dns_get_record($h, DNS_CNAME + DNS_MX)) {
 502                         return true;
 503                 }
 504                 return false;
 505         }
 506
 507         /**
 508          * @brief Check if URL is allowed
 509          *
 510          * Check $url against our list of allowed sites,
 511          * wildcards allowed. If allowed_sites is unset return true;
 512          *
 513          * @param string $url URL which get tested
 514          * @return boolean True if url is allowed otherwise return false
 515          */
 516         public static function isUrlAllowed($url)
 517         {
 518                 $h = @parse_url($url);
 519
 520                 if (! $h) {
 521                         return false;
 522                 }
 523
 524                 $str_allowed = Config::get('system', 'allowed_sites');
 525                 if (! $str_allowed) {
 526                         return true;
 527                 }
 528
 529                 $found = false;
 530
 531                 $host = strtolower($h['host']);
 532
 533                 // always allow our own site
 534                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 535                         return true;
 536                 }
 537
 538                 $fnmatch = function_exists('fnmatch');
 539                 $allowed = explode(',', $str_allowed);
 540
 541                 if (count($allowed)) {
 542                         foreach ($allowed as $a) {
 543                                 $pat = strtolower(trim($a));
 544                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 545                                         $found = true;
 546                                         break;
 547                                 }
 548                         }
 549                 }
 550                 return $found;
 551         }
 552
 553         /**
 554          * Checks if the provided url domain is on the domain blocklist.
 555          * Returns true if it is or malformed URL, false if not.
 556          *
 557          * @param string $url The url to check the domain from
 558          *
 559          * @return boolean
 560          */
 561         public static function isUrlBlocked($url)
 562         {
 563                 $host = @parse_url($url, PHP_URL_HOST);
 564                 if (!$host) {
 565                         return false;
 566                 }
 567
 568                 $domain_blocklist = Config::get('system', 'blocklist', []);
 569                 if (!$domain_blocklist) {
 570                         return false;
 571                 }
 572
 573                 foreach ($domain_blocklist as $domain_block) {
 574                         if (strcasecmp($domain_block['domain'], $host) === 0) {
 575                                 return true;
 576                         }
 577                 }
 578
 579                 return false;
 580         }
 581
 582         /**
 583          * @brief Check if email address is allowed to register here.
 584          *
 585          * Compare against our list (wildcards allowed).
 586          *
 587          * @param  string $email email address
 588          * @return boolean False if not allowed, true if allowed
 589          *    or if allowed list is not configured
 590          */
 591         public static function isEmailDomainAllowed($email)
 592         {
 593                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 594                 if (!$domain) {
 595                         return false;
 596                 }
 597
 598                 $str_allowed = Config::get('system', 'allowed_email', '');
 599                 if (!x($str_allowed)) {
 600                         return true;
 601                 }
 602
 603                 $allowed = explode(',', $str_allowed);
 604
 605                 return self::isDomainAllowed($domain, $allowed);
 606         }
 607
 608         /**
 609          * Checks for the existence of a domain in a domain list
 610          *
 611          * @brief Checks for the existence of a domain in a domain list
 612          * @param string $domain
 613          * @param array  $domain_list
 614          * @return boolean
 615          */
 616         public static function isDomainAllowed($domain, array $domain_list)
 617         {
 618                 $found = false;
 619
 620                 foreach ($domain_list as $item) {
 621                         $pat = strtolower(trim($item));
 622                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 623                                 $found = true;
 624                                 break;
 625                         }
 626                 }
 627
 628                 return $found;
 629         }
 630
 631         public static function lookupAvatarByEmail($email)
 632         {
 633                 $avatar['size'] = 175;
 634                 $avatar['email'] = $email;
 635                 $avatar['url'] = '';
 636                 $avatar['success'] = false;
 637
 638                 Addon::callHooks('avatar_lookup', $avatar);
 639
 640                 if (! $avatar['success']) {
 641                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 642                 }
 643
 644                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 645                 return $avatar['url'];
 646         }
 647
 648         /**
 649          * @brief Remove Google Analytics and other tracking platforms params from URL
 650          *
 651          * @param string $url Any user-submitted URL that may contain tracking params
 652          * @return string The same URL stripped of tracking parameters
 653          */
 654         public static function stripTrackingQueryParams($url)
 655         {
 656                 $urldata = parse_url($url);
 657                 if (!empty($urldata["query"])) {
 658                         $query = $urldata["query"];
 659                         parse_str($query, $querydata);
 660
 661                         if (is_array($querydata)) {
 662                                 foreach ($querydata as $param => $value) {
 663                                         if (in_array(
 664                                                 $param,
 665                                                 [
 666                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 667                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 668                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 669                                                         "awesm", "wtrid",
 670                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 671                                                 )
 672                                         ) {
 673                                                 $pair = $param . "=" . urlencode($value);
 674                                                 $url = str_replace($pair, "", $url);
 675
 676                                                 // Second try: if the url isn't encoded completely
 677                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 678                                                 $url = str_replace($pair, "", $url);
 679
 680                                                 // Third try: Maybey the url isn't encoded at all
 681                                                 $pair = $param . "=" . $value;
 682                                                 $url = str_replace($pair, "", $url);
 683
 684                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 685                                         }
 686                                 }
 687                         }
 688
 689                         if (substr($url, -1, 1) == "?") {
 690                                 $url = substr($url, 0, -1);
 691                         }
 692                 }
 693
 694                 return $url;
 695         }
 696
 697         /**
 698          * @brief Returns the original URL of the provided URL
 699          *
 700          * This function strips tracking query params and follows redirections, either
 701          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 702          *
 703          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 704          *
 705          * @see ParseUrl::getSiteinfo
 706          *
 707          * @param string $url       A user-submitted URL
 708          * @param int    $depth     The current redirection recursion level (internal)
 709          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 710          * @return string A canonical URL
 711          */
 712         public static function finalUrl($url, $depth = 1, $fetchbody = false)
 713         {
 714                 $a = get_app();
 715
 716                 $url = self::stripTrackingQueryParams($url);
 717
 718                 if ($depth > 10) {
 719                         return $url;
 720                 }
 721
 722                 $url = trim($url, "'");
 723
 724                 $stamp1 = microtime(true);
 725
 726                 $ch = curl_init();
 727                 curl_setopt($ch, CURLOPT_URL, $url);
 728                 curl_setopt($ch, CURLOPT_HEADER, 1);
 729                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 730                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 731                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 732                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 733
 734                 curl_exec($ch);
 735                 $curl_info = @curl_getinfo($ch);
 736                 $http_code = $curl_info['http_code'];
 737                 curl_close($ch);
 738
 739                 $a->save_timestamp($stamp1, "network");
 740
 741                 if ($http_code == 0) {
 742                         return $url;
 743                 }
 744
 745                 if (in_array($http_code, ['301', '302'])) {
 746                         if (!empty($curl_info['redirect_url'])) {
 747                                 return self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 748                         } elseif (!empty($curl_info['location'])) {
 749                                 return self::finalUrl($curl_info['location'], ++$depth, $fetchbody);
 750                         }
 751                 }
 752
 753                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 754                 if (!$fetchbody) {
 755                         return(self::finalUrl($url, ++$depth, true));
 756                 }
 757
 758                 // if the file is too large then exit
 759                 if ($curl_info["download_content_length"] > 1000000) {
 760                         return $url;
 761                 }
 762
 763                 // if it isn't a HTML file then exit
 764                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 765                         return $url;
 766                 }
 767
 768                 $stamp1 = microtime(true);
 769
 770                 $ch = curl_init();
 771                 curl_setopt($ch, CURLOPT_URL, $url);
 772                 curl_setopt($ch, CURLOPT_HEADER, 0);
 773                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 774                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 775                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 776                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 777
 778                 $body = curl_exec($ch);
 779                 curl_close($ch);
 780
 781                 $a->save_timestamp($stamp1, "network");
 782
 783                 if (trim($body) == "") {
 784                         return $url;
 785                 }
 786
 787                 // Check for redirect in meta elements
 788                 $doc = new DOMDocument();
 789                 @$doc->loadHTML($body);
 790
 791                 $xpath = new DomXPath($doc);
 792
 793                 $list = $xpath->query("//meta[@content]");
 794                 foreach ($list as $node) {
 795                         $attr = [];
 796                         if ($node->attributes->length) {
 797                                 foreach ($node->attributes as $attribute) {
 798                                         $attr[$attribute->name] = $attribute->value;
 799                                 }
 800                         }
 801
 802                         if (@$attr["http-equiv"] == 'refresh') {
 803                                 $path = $attr["content"];
 804                                 $pathinfo = explode(";", $path);
 805                                 foreach ($pathinfo as $value) {
 806                                         if (substr(strtolower($value), 0, 4) == "url=") {
 807                                                 return self::finalUrl(substr($value, 4), ++$depth);
 808                                         }
 809                                 }
 810                         }
 811                 }
 812
 813                 return $url;
 814         }
 815
 816         /**
 817          * @brief Find the matching part between two url
 818          *
 819          * @param string $url1
 820          * @param string $url2
 821          * @return string The matching part
 822          */
 823         public static function getUrlMatch($url1, $url2)
 824         {
 825                 if (($url1 == "") || ($url2 == "")) {
 826                         return "";
 827                 }
 828
 829                 $url1 = normalise_link($url1);
 830                 $url2 = normalise_link($url2);
 831
 832                 $parts1 = parse_url($url1);
 833                 $parts2 = parse_url($url2);
 834
 835                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 836                         return "";
 837                 }
 838
 839                 if (empty($parts1["scheme"])) {
 840                         $parts1["scheme"] = '';
 841                 }
 842                 if (empty($parts2["scheme"])) {
 843                         $parts2["scheme"] = '';
 844                 }
 845
 846                 if ($parts1["scheme"] != $parts2["scheme"]) {
 847                         return "";
 848                 }
 849
 850                 if (empty($parts1["host"])) {
 851                         $parts1["host"] = '';
 852                 }
 853                 if (empty($parts2["host"])) {
 854                         $parts2["host"] = '';
 855                 }
 856
 857                 if ($parts1["host"] != $parts2["host"]) {
 858                         return "";
 859                 }
 860
 861                 if (empty($parts1["port"])) {
 862                         $parts1["port"] = '';
 863                 }
 864                 if (empty($parts2["port"])) {
 865                         $parts2["port"] = '';
 866                 }
 867
 868                 if ($parts1["port"] != $parts2["port"]) {
 869                         return "";
 870                 }
 871
 872                 $match = $parts1["scheme"]."://".$parts1["host"];
 873
 874                 if ($parts1["port"]) {
 875                         $match .= ":".$parts1["port"];
 876                 }
 877
 878                 if (empty($parts1["path"])) {
 879                         $parts1["path"] = '';
 880                 }
 881                 if (empty($parts2["path"])) {
 882                         $parts2["path"] = '';
 883                 }
 884
 885                 $pathparts1 = explode("/", $parts1["path"]);
 886                 $pathparts2 = explode("/", $parts2["path"]);
 887
 888                 $i = 0;
 889                 $path = "";
 890                 do {
 891                         $path1 = defaults($pathparts1, $i, '');
 892                         $path2 = defaults($pathparts2, $i, '');
 893
 894                         if ($path1 == $path2) {
 895                                 $path .= $path1."/";
 896                         }
 897                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 898
 899                 $match .= $path;
 900
 901                 return normalise_link($match);
 902         }
 903
 904         /**
 905          * @brief Glue url parts together
 906          *
 907          * @param array $parsed URL parts
 908          *
 909          * @return string The glued URL
 910          */
 911         public static function unparseURL($parsed)
 912         {
 913                 $get = function ($key) use ($parsed) {
 914                         return isset($parsed[$key]) ? $parsed[$key] : null;
 915                 };
 916
 917                 $pass      = $get('pass');
 918                 $user      = $get('user');
 919                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 920                 $port      = $get('port');
 921                 $scheme    = $get('scheme');
 922                 $query     = $get('query');
 923                 $fragment  = $get('fragment');
 924                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 925                                                 $get('host') .
 926                                                 ($port ? ":$port" : '');
 927
 928                 return  (strlen($scheme) ? $scheme.":" : '') .
 929                         (strlen($authority) ? "//".$authority : '') .
 930                         $get('path') .
 931                         (strlen($query) ? "?".$query : '') .
 932                         (strlen($fragment) ? "#".$fragment : '');
 933         }
 934 }