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