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                 $ret = self::fetchUrlFull($url, $binary, $redirects, $timeout, $accept_content, $cookiejar);
  40
  41                 return $ret['body'];
  42         }
  43
  44         /**
  45          * @brief Curl wrapper with array of return values.
  46          *
  47          * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
  48          * all the information collected during the fetch.
  49          *
  50          * @param string  $url            URL to fetch
  51          * @param boolean $binary         default false
  52          *                                TRUE if asked to return binary results (file download)
  53          * @param integer $redirects      The recursion counter for internal use - default 0
  54          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  55          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  56          * @param string  $cookiejar      Path to cookie jar file
  57          *
  58          * @return array With all relevant information, 'body' contains the actual fetched content.
  59          */
  60         public static function fetchUrlFull($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  61         {
  62                 return self::curl(
  63                         $url,
  64                         $binary,
  65                         $redirects,
  66                         ['timeout'=>$timeout,
  67                         'accept_content'=>$accept_content,
  68                         'cookiejar'=>$cookiejar
  69                         ]
  70                 );
  71         }
  72
  73         /**
  74          * @brief fetches an URL.
  75          *
  76          * @param string  $url       URL to fetch
  77          * @param boolean $binary    default false
  78          *                           TRUE if asked to return binary results (file download)
  79          * @param int     $redirects The recursion counter for internal use - default 0
  80          * @param array   $opts      (optional parameters) assoziative array with:
  81          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  82          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  83          *                           'http_auth' => username:password
  84          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  85          *                           'nobody' => only return the header
  86          *                           'cookiejar' => path to cookie jar file
  87          *
  88          * @return array an assoziative array with:
  89          *    int 'return_code' => HTTP return code or 0 if timeout or failure
  90          *    boolean 'success' => boolean true (if HTTP 2xx result) or false
  91          *    string 'redirect_url' => in case of redirect, content was finally retrieved from this URL
  92          *    string 'header' => HTTP headers
  93          *    string 'body' => fetched content
  94          */
  95         public static function curl($url, $binary = false, &$redirects = 0, $opts = [])
  96         {
  97                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
  98
  99                 $stamp1 = microtime(true);
 100
 101                 $a = get_app();
 102
 103                 $parts = parse_url($url);
 104                 $path_parts = explode('/', $parts['path']);
 105                 foreach ($path_parts as $part) {
 106                         if (strlen($part) <> mb_strlen($part)) {
 107                                 $parts2[] = rawurlencode($part);
 108                         } else {
 109                                 $parts2[] = $part;
 110                         }
 111                 }
 112                 $parts['path'] =  implode('/', $parts2);
 113                 $url = self::unparseURL($parts);
 114
 115                 if (self::isUrlBlocked($url)) {
 116                         logger('domain of ' . $url . ' is blocked', LOGGER_DATA);
 117                         return $ret;
 118                 }
 119
 120                 $ch = @curl_init($url);
 121
 122                 if (($redirects > 8) || (!$ch)) {
 123                         return $ret;
 124                 }
 125
 126                 @curl_setopt($ch, CURLOPT_HEADER, true);
 127
 128                 if (x($opts, "cookiejar")) {
 129                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 130                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 131                 }
 132
 133                 // These settings aren't needed. We're following the location already.
 134                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 135                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 136
 137                 if (x($opts, 'accept_content')) {
 138                         curl_setopt(
 139                                 $ch,
 140                                 CURLOPT_HTTPHEADER,
 141                                 ['Accept: ' . $opts['accept_content']]
 142                         );
 143                 }
 144
 145                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 146                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 147
 148                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 149
 150                 if ($range > 0) {
 151                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 152                 }
 153
 154                 // Without this setting it seems as if some webservers send compressed content
 155                 // This seems to confuse curl so that it shows this uncompressed.
 156                 /// @todo  We could possibly set this value to "gzip" or something similar
 157                 curl_setopt($ch, CURLOPT_ENCODING, '');
 158
 159                 if (x($opts, 'headers')) {
 160                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 161                 }
 162
 163                 if (x($opts, 'nobody')) {
 164                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 165                 }
 166
 167                 if (x($opts, 'timeout')) {
 168                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 169                 } else {
 170                         $curl_time = Config::get('system', 'curl_timeout', 60);
 171                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 172                 }
 173
 174                 // by default we will allow self-signed certs
 175                 // but you can override this
 176
 177                 $check_cert = Config::get('system', 'verifyssl');
 178                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 179
 180                 if ($check_cert) {
 181                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 182                 }
 183
 184                 $proxy = Config::get('system', 'proxy');
 185
 186                 if (strlen($proxy)) {
 187                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 188                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 189                         $proxyuser = @Config::get('system', 'proxyuser');
 190
 191                         if (strlen($proxyuser)) {
 192                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 193                         }
 194                 }
 195
 196                 if (Config::get('system', 'ipv4_resolve', false)) {
 197                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 198                 }
 199
 200                 if ($binary) {
 201                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 202                 }
 203
 204                 $a->set_curl_code(0);
 205
 206                 // don't let curl abort the entire application
 207                 // if it throws any errors.
 208
 209                 $s = @curl_exec($ch);
 210                 $curl_info = @curl_getinfo($ch);
 211
 212                 // Special treatment for HTTP Code 416
 213                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 214                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 215                         @curl_setopt($ch, CURLOPT_RANGE, '');
 216                         $s = @curl_exec($ch);
 217                         $curl_info = @curl_getinfo($ch);
 218                 }
 219
 220                 if (curl_errno($ch) !== CURLE_OK) {
 221                         logger('error fetching ' . $url . ': ' . curl_error($ch), LOGGER_NORMAL);
 222                 }
 223
 224                 $ret['errno'] = curl_errno($ch);
 225
 226                 $base = $s;
 227                 $ret['info'] = $curl_info;
 228
 229                 $http_code = $curl_info['http_code'];
 230
 231                 logger($url . ': ' . $http_code . " " . $s, LOGGER_DATA);
 232                 $header = '';
 233
 234                 // Pull out multiple headers, e.g. proxy and continuation headers
 235                 // allow for HTTP/2.x without fixing code
 236
 237                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 238                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 239                         $header .= $chunk;
 240                         $base = substr($base, strlen($chunk));
 241                 }
 242
 243                 $a->set_curl_code($http_code);
 244                 $a->set_curl_content_type($curl_info['content_type']);
 245                 $a->set_curl_headers($header);
 246
 247                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 248                         $new_location_info = @parse_url($curl_info['redirect_url']);
 249                         $old_location_info = @parse_url($curl_info['url']);
 250
 251                         $newurl = $curl_info['redirect_url'];
 252
 253                         if (($new_location_info['path'] == '') && ($new_location_info['host'] != '')) {
 254                                 $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
 255                         }
 256
 257                         $matches = [];
 258
 259                         if (preg_match('/(Location:|URI:)(.*?)\n/i', $header, $matches)) {
 260                                 $newurl = trim(array_pop($matches));
 261                         }
 262                         if (strpos($newurl, '/') === 0) {
 263                                 $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
 264                         }
 265                         $old_location_query = @parse_url($url, PHP_URL_QUERY);
 266
 267                         if ($old_location_query != '') {
 268                                 $newurl .= '?' . $old_location_query;
 269                         }
 270
 271                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 272                                 $redirects++;
 273                                 @curl_close($ch);
 274                                 return self::curl($newurl, $binary, $redirects, $opts);
 275                         }
 276                 }
 277
 278                 $a->set_curl_code($http_code);
 279                 $a->set_curl_content_type($curl_info['content_type']);
 280
 281                 $rc = intval($http_code);
 282                 $ret['return_code'] = $rc;
 283                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 284                 $ret['redirect_url'] = $url;
 285
 286                 if (!$ret['success']) {
 287                         $ret['error'] = curl_error($ch);
 288                         $ret['debug'] = $curl_info;
 289                         logger('error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 290                         logger('debug: '.print_r($curl_info, true), LOGGER_DATA);
 291                 }
 292
 293                 $ret['body'] = substr($s, strlen($header));
 294                 $ret['header'] = $header;
 295
 296                 if (x($opts, 'debug')) {
 297                         $ret['debug'] = $curl_info;
 298                 }
 299
 300                 @curl_close($ch);
 301
 302                 $a->save_timestamp($stamp1, 'network');
 303
 304                 return($ret);
 305         }
 306
 307         /**
 308          * @brief Send POST request to $url
 309          *
 310          * @param string  $url       URL to post
 311          * @param mixed   $params    array of POST variables
 312          * @param string  $headers   HTTP headers
 313          * @param integer $redirects Recursion counter for internal use - default = 0
 314          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 315          *
 316          * @return string The content
 317          */
 318         public static function post($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 319         {
 320                 $stamp1 = microtime(true);
 321
 322                 if (self::isUrlBlocked($url)) {
 323                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 324                         return false;
 325                 }
 326
 327                 $a = get_app();
 328                 $ch = curl_init($url);
 329
 330                 if (($redirects > 8) || (!$ch)) {
 331                         return false;
 332                 }
 333
 334                 logger('post_url: start ' . $url, LOGGER_DATA);
 335
 336                 curl_setopt($ch, CURLOPT_HEADER, true);
 337                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 338                 curl_setopt($ch, CURLOPT_POST, 1);
 339                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 340                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 341
 342                 if (Config::get('system', 'ipv4_resolve', false)) {
 343                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 344                 }
 345
 346                 if (intval($timeout)) {
 347                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 348                 } else {
 349                         $curl_time = Config::get('system', 'curl_timeout', 60);
 350                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 351                 }
 352
 353                 if (defined('LIGHTTPD')) {
 354                         if (!is_array($headers)) {
 355                                 $headers = ['Expect:'];
 356                         } else {
 357                                 if (!in_array('Expect:', $headers)) {
 358                                         array_push($headers, 'Expect:');
 359                                 }
 360                         }
 361                 }
 362
 363                 if ($headers) {
 364                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 365                 }
 366
 367                 $check_cert = Config::get('system', 'verifyssl');
 368                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 369
 370                 if ($check_cert) {
 371                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 372                 }
 373
 374                 $proxy = Config::get('system', 'proxy');
 375
 376                 if (strlen($proxy)) {
 377                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 378                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 379                         $proxyuser = Config::get('system', 'proxyuser');
 380                         if (strlen($proxyuser)) {
 381                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 382                         }
 383                 }
 384
 385                 $a->set_curl_code(0);
 386
 387                 // don't let curl abort the entire application
 388                 // if it throws any errors.
 389
 390                 $s = @curl_exec($ch);
 391
 392                 $base = $s;
 393                 $curl_info = curl_getinfo($ch);
 394                 $http_code = $curl_info['http_code'];
 395
 396                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 397
 398                 $header = '';
 399
 400                 // Pull out multiple headers, e.g. proxy and continuation headers
 401                 // allow for HTTP/2.x without fixing code
 402
 403                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 404                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 405                         $header .= $chunk;
 406                         $base = substr($base, strlen($chunk));
 407                 }
 408
 409                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 410                         $matches = [];
 411                         $new_location_info = @parse_url($curl_info['redirect_url']);
 412                         $old_location_info = @parse_url($curl_info['url']);
 413
 414                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 415                         $newurl = trim(array_pop($matches));
 416
 417                         if (strpos($newurl, '/') === 0) {
 418                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 419                         }
 420
 421                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 422                                 $redirects++;
 423                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 424                                 return self::post($newurl, $params, $headers, $redirects, $timeout);
 425                         }
 426                 }
 427
 428                 $a->set_curl_code($http_code);
 429
 430                 $body = substr($s, strlen($header));
 431
 432                 $a->set_curl_headers($header);
 433
 434                 curl_close($ch);
 435
 436                 $a->save_timestamp($stamp1, 'network');
 437
 438                 logger('post_url: end ' . $url, LOGGER_DATA);
 439
 440                 return $body;
 441         }
 442
 443         /**
 444          * @brief Check URL to see if it's real
 445          *
 446          * Take a URL from the wild, prepend http:// if necessary
 447          * and check DNS to see if it's real (or check if is a valid IP address)
 448          *
 449          * @param string $url The URL to be validated
 450          * @return string|boolean The actual working URL, false else
 451          */
 452         public static function isUrlValid($url)
 453         {
 454                 if (Config::get('system', 'disable_url_validation')) {
 455                         return $url;
 456                 }
 457
 458                 // no naked subdomains (allow localhost for tests)
 459                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 460                         return false;
 461                 }
 462
 463                 if (substr($url, 0, 4) != 'http') {
 464                         $url = 'http://' . $url;
 465                 }
 466
 467                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 468                 $h = @parse_url($url);
 469
 470                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 471                         return $url;
 472                 }
 473
 474                 return false;
 475         }
 476
 477         /**
 478          * @brief Checks that email is an actual resolvable internet address
 479          *
 480          * @param string $addr The email address
 481          * @return boolean True if it's a valid email address, false if it's not
 482          */
 483         public static function isEmailDomainValid($addr)
 484         {
 485                 if (Config::get('system', 'disable_email_validation')) {
 486                         return true;
 487                 }
 488
 489                 if (! strpos($addr, '@')) {
 490                         return false;
 491                 }
 492
 493                 $h = substr($addr, strpos($addr, '@') + 1);
 494
 495                 if (($h) && (dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 496                         return true;
 497                 }
 498                 if (($h) && dns_get_record($h, DNS_CNAME + DNS_MX)) {
 499                         return true;
 500                 }
 501                 return false;
 502         }
 503
 504         /**
 505          * @brief Check if URL is allowed
 506          *
 507          * Check $url against our list of allowed sites,
 508          * wildcards allowed. If allowed_sites is unset return true;
 509          *
 510          * @param string $url URL which get tested
 511          * @return boolean True if url is allowed otherwise return false
 512          */
 513         public static function isUrlAllowed($url)
 514         {
 515                 $h = @parse_url($url);
 516
 517                 if (! $h) {
 518                         return false;
 519                 }
 520
 521                 $str_allowed = Config::get('system', 'allowed_sites');
 522                 if (! $str_allowed) {
 523                         return true;
 524                 }
 525
 526                 $found = false;
 527
 528                 $host = strtolower($h['host']);
 529
 530                 // always allow our own site
 531                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 532                         return true;
 533                 }
 534
 535                 $fnmatch = function_exists('fnmatch');
 536                 $allowed = explode(',', $str_allowed);
 537
 538                 if (count($allowed)) {
 539                         foreach ($allowed as $a) {
 540                                 $pat = strtolower(trim($a));
 541                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 542                                         $found = true;
 543                                         break;
 544                                 }
 545                         }
 546                 }
 547                 return $found;
 548         }
 549
 550         /**
 551          * Checks if the provided url domain is on the domain blocklist.
 552          * Returns true if it is or malformed URL, false if not.
 553          *
 554          * @param string $url The url to check the domain from
 555          *
 556          * @return boolean
 557          */
 558         public static function isUrlBlocked($url)
 559         {
 560                 $h = @parse_url($url);
 561
 562                 if (! $h) {
 563                         return true;
 564                 }
 565
 566                 $domain_blocklist = Config::get('system', 'blocklist', []);
 567                 if (! $domain_blocklist) {
 568                         return false;
 569                 }
 570
 571                 $host = strtolower($h['host']);
 572
 573                 foreach ($domain_blocklist as $domain_block) {
 574                         if (strtolower($domain_block['domain']) == $host) {
 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 (is_string($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 ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 746                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 747                 ) {
 748                         if ($curl_info['redirect_url'] != "") {
 749                                 return(self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody));
 750                         } else {
 751                                 return(self::finalUrl($curl_info['location'], ++$depth, $fetchbody));
 752                         }
 753                 }
 754
 755                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 756                 if (!$fetchbody) {
 757                         return(self::finalUrl($url, ++$depth, true));
 758                 }
 759
 760                 // if the file is too large then exit
 761                 if ($curl_info["download_content_length"] > 1000000) {
 762                         return($url);
 763                 }
 764
 765                 // if it isn't a HTML file then exit
 766                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 767                         return($url);
 768                 }
 769
 770                 $stamp1 = microtime(true);
 771
 772                 $ch = curl_init();
 773                 curl_setopt($ch, CURLOPT_URL, $url);
 774                 curl_setopt($ch, CURLOPT_HEADER, 0);
 775                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 776                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 777                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 778                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 779
 780                 $body = curl_exec($ch);
 781                 curl_close($ch);
 782
 783                 $a->save_timestamp($stamp1, "network");
 784
 785                 if (trim($body) == "") {
 786                         return($url);
 787                 }
 788
 789                 // Check for redirect in meta elements
 790                 $doc = new DOMDocument();
 791                 @$doc->loadHTML($body);
 792
 793                 $xpath = new DomXPath($doc);
 794
 795                 $list = $xpath->query("//meta[@content]");
 796                 foreach ($list as $node) {
 797                         $attr = [];
 798                         if ($node->attributes->length) {
 799                                 foreach ($node->attributes as $attribute) {
 800                                         $attr[$attribute->name] = $attribute->value;
 801                                 }
 802                         }
 803
 804                         if (@$attr["http-equiv"] == 'refresh') {
 805                                 $path = $attr["content"];
 806                                 $pathinfo = explode(";", $path);
 807                                 foreach ($pathinfo as $value) {
 808                                         if (substr(strtolower($value), 0, 4) == "url=") {
 809                                                 return(self::finalUrl(substr($value, 4), ++$depth));
 810                                         }
 811                                 }
 812                         }
 813                 }
 814
 815                 return $url;
 816         }
 817
 818         /**
 819          * @brief Find the matching part between two url
 820          *
 821          * @param string $url1
 822          * @param string $url2
 823          * @return string The matching part
 824          */
 825         public static function getUrlMatch($url1, $url2)
 826         {
 827                 if (($url1 == "") || ($url2 == "")) {
 828                         return "";
 829                 }
 830
 831                 $url1 = normalise_link($url1);
 832                 $url2 = normalise_link($url2);
 833
 834                 $parts1 = parse_url($url1);
 835                 $parts2 = parse_url($url2);
 836
 837                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 838                         return "";
 839                 }
 840
 841                 if ($parts1["scheme"] != $parts2["scheme"]) {
 842                         return "";
 843                 }
 844
 845                 if ($parts1["host"] != $parts2["host"]) {
 846                         return "";
 847                 }
 848
 849                 if ($parts1["port"] != $parts2["port"]) {
 850                         return "";
 851                 }
 852
 853                 $match = $parts1["scheme"]."://".$parts1["host"];
 854
 855                 if ($parts1["port"]) {
 856                         $match .= ":".$parts1["port"];
 857                 }
 858
 859                 $pathparts1 = explode("/", $parts1["path"]);
 860                 $pathparts2 = explode("/", $parts2["path"]);
 861
 862                 $i = 0;
 863                 $path = "";
 864                 do {
 865                         $path1 = $pathparts1[$i];
 866                         $path2 = $pathparts2[$i];
 867
 868                         if ($path1 == $path2) {
 869                                 $path .= $path1."/";
 870                         }
 871                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 872
 873                 $match .= $path;
 874
 875                 return normalise_link($match);
 876         }
 877
 878         /**
 879          * @brief Glue url parts together
 880          *
 881          * @param array $parsed URL parts
 882          *
 883          * @return string The glued URL
 884          */
 885         public static function unparseURL($parsed)
 886         {
 887                 $get = function ($key) use ($parsed) {
 888                         return isset($parsed[$key]) ? $parsed[$key] : null;
 889                 };
 890
 891                 $pass      = $get('pass');
 892                 $user      = $get('user');
 893                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 894                 $port      = $get('port');
 895                 $scheme    = $get('scheme');
 896                 $query     = $get('query');
 897                 $fragment  = $get('fragment');
 898                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 899                                                 $get('host') .
 900                                                 ($port ? ":$port" : '');
 901
 902                 return  (strlen($scheme) ? $scheme.":" : '') .
 903                         (strlen($authority) ? "//".$authority : '') .
 904                         $get('path') .
 905                         (strlen($query) ? "?".$query : '') .
 906                         (strlen($fragment) ? "#".$fragment : '');
 907         }
 908 }