src/Util/Network.php

   1 <?php
   2 /**
   3  * @file src/Util/Network.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use DOMDocument;
   8 use DomXPath;
   9 use Friendica\Core\Hook;
  10 use Friendica\Core\Logger;
  11 use Friendica\Core\System;
  12 use Friendica\DI;
  13 use Friendica\Network\CurlResult;
  14
  15 class Network
  16 {
  17         /**
  18          * Curl wrapper
  19          *
  20          * If binary flag is true, return binary results.
  21          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
  22          * to preserve cookies from one request to the next.
  23          *
  24          * @param string  $url            URL to fetch
  25          * @param bool    $binary         default false
  26          *                                TRUE if asked to return binary results (file download)
  27          * @param int     $timeout        Timeout in seconds, default system config value or 60 seconds
  28          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  29          * @param string  $cookiejar      Path to cookie jar file
  30          * @param int     $redirects      The recursion counter for internal use - default 0
  31          *
  32          * @return string The fetched content
  33          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  34          */
  35         public static function fetchUrl(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
  36         {
  37                 $ret = self::fetchUrlFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
  38
  39                 return $ret->getBody();
  40         }
  41
  42         /**
  43          * Curl wrapper with array of return values.
  44          *
  45          * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
  46          * all the information collected during the fetch.
  47          *
  48          * @param string  $url            URL to fetch
  49          * @param bool    $binary         default false
  50          *                                TRUE if asked to return binary results (file download)
  51          * @param int     $timeout        Timeout in seconds, default system config value or 60 seconds
  52          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  53          * @param string  $cookiejar      Path to cookie jar file
  54          * @param int     $redirects      The recursion counter for internal use - default 0
  55          *
  56          * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
  57          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  58          */
  59         public static function fetchUrlFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
  60         {
  61                 return self::curl(
  62                         $url,
  63                         $binary,
  64                         [
  65                                 'timeout'        => $timeout,
  66                                 'accept_content' => $accept_content,
  67                                 'cookiejar'      => $cookiejar
  68                         ],
  69                         $redirects
  70                 );
  71         }
  72
  73         /**
  74          * fetches an URL.
  75          *
  76          * @param string  $url       URL to fetch
  77          * @param bool    $binary    default false
  78          *                           TRUE if asked to return binary results (file download)
  79          * @param array   $opts      (optional parameters) assoziative array with:
  80          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  81          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  82          *                           'http_auth' => username:password
  83          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  84          *                           'nobody' => only return the header
  85          *                           'cookiejar' => path to cookie jar file
  86          *                           'header' => header array
  87          * @param int     $redirects The recursion counter for internal use - default 0
  88          *
  89          * @return CurlResult
  90          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  91          */
  92         public static function curl(string $url, bool $binary = false, array $opts = [], int &$redirects = 0)
  93         {
  94                 $stamp1 = microtime(true);
  95
  96                 $a = DI::app();
  97
  98                 if (strlen($url) > 1000) {
  99                         Logger::log('URL is longer than 1000 characters. Callstack: ' . System::callstack(20), Logger::DEBUG);
 100                         return CurlResult::createErrorCurl(substr($url, 0, 200));
 101                 }
 102
 103                 $parts2 = [];
 104                 $parts = parse_url($url);
 105                 $path_parts = explode('/', $parts['path'] ?? '');
 106                 foreach ($path_parts as $part) {
 107                         if (strlen($part) <> mb_strlen($part)) {
 108                                 $parts2[] = rawurlencode($part);
 109                         } else {
 110                                 $parts2[] = $part;
 111                         }
 112                 }
 113                 $parts['path'] = implode('/', $parts2);
 114                 $url = self::unparseURL($parts);
 115
 116                 if (self::isUrlBlocked($url)) {
 117                         Logger::log('domain of ' . $url . ' is blocked', Logger::DATA);
 118                         return CurlResult::createErrorCurl($url);
 119                 }
 120
 121                 $ch = @curl_init($url);
 122
 123                 if (($redirects > 8) || (!$ch)) {
 124                         return CurlResult::createErrorCurl($url);
 125                 }
 126
 127                 @curl_setopt($ch, CURLOPT_HEADER, true);
 128
 129                 if (!empty($opts['cookiejar'])) {
 130                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 131                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 132                 }
 133
 134                 // These settings aren't needed. We're following the location already.
 135                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 136                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 137
 138                 if (!empty($opts['accept_content'])) {
 139                         curl_setopt(
 140                                 $ch,
 141                                 CURLOPT_HTTPHEADER,
 142                                 ['Accept: ' . $opts['accept_content']]
 143                         );
 144                 }
 145
 146                 if (!empty($opts['header'])) {
 147                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 148                 }
 149
 150                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 151                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 152
 153                 $range = intval(DI::config()->get('system', 'curl_range_bytes', 0));
 154
 155                 if ($range > 0) {
 156                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 157                 }
 158
 159                 // Without this setting it seems as if some webservers send compressed content
 160                 // This seems to confuse curl so that it shows this uncompressed.
 161                 /// @todo  We could possibly set this value to "gzip" or something similar
 162                 curl_setopt($ch, CURLOPT_ENCODING, '');
 163
 164                 if (!empty($opts['headers'])) {
 165                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 166                 }
 167
 168                 if (!empty($opts['nobody'])) {
 169                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 170                 }
 171
 172                 if (!empty($opts['timeout'])) {
 173                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 174                 } else {
 175                         $curl_time = DI::config()->get('system', 'curl_timeout', 60);
 176                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 177                 }
 178
 179                 // by default we will allow self-signed certs
 180                 // but you can override this
 181
 182                 $check_cert = DI::config()->get('system', 'verifyssl');
 183                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 184
 185                 if ($check_cert) {
 186                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 187                 }
 188
 189                 $proxy = DI::config()->get('system', 'proxy');
 190
 191                 if (strlen($proxy)) {
 192                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 193                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 194                         $proxyuser = @DI::config()->get('system', 'proxyuser');
 195
 196                         if (strlen($proxyuser)) {
 197                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 198                         }
 199                 }
 200
 201                 if (DI::config()->get('system', 'ipv4_resolve', false)) {
 202                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 203                 }
 204
 205                 if ($binary) {
 206                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 207                 }
 208
 209                 // don't let curl abort the entire application
 210                 // if it throws any errors.
 211
 212                 $s = @curl_exec($ch);
 213                 $curl_info = @curl_getinfo($ch);
 214
 215                 // Special treatment for HTTP Code 416
 216                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 217                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 218                         @curl_setopt($ch, CURLOPT_RANGE, '');
 219                         $s = @curl_exec($ch);
 220                         $curl_info = @curl_getinfo($ch);
 221                 }
 222
 223                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 224
 225                 if ($curlResponse->isRedirectUrl()) {
 226                         $redirects++;
 227                         Logger::log('curl: redirect ' . $url . ' to ' . $curlResponse->getRedirectUrl());
 228                         @curl_close($ch);
 229                         return self::curl($curlResponse->getRedirectUrl(), $binary, $opts, $redirects);
 230                 }
 231
 232                 @curl_close($ch);
 233
 234                 DI::profiler()->saveTimestamp($stamp1, 'network', System::callstack());
 235
 236                 return $curlResponse;
 237         }
 238
 239         /**
 240          * Send POST request to $url
 241          *
 242          * @param string  $url       URL to post
 243          * @param mixed   $params    array of POST variables
 244          * @param array   $headers   HTTP headers
 245          * @param int     $redirects Recursion counter for internal use - default = 0
 246          * @param int     $timeout   The timeout in seconds, default system config value or 60 seconds
 247          *
 248          * @return CurlResult The content
 249          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 250          */
 251         public static function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 252         {
 253                 $stamp1 = microtime(true);
 254
 255                 if (self::isUrlBlocked($url)) {
 256                         Logger::log('post_url: domain of ' . $url . ' is blocked', Logger::DATA);
 257                         return CurlResult::createErrorCurl($url);
 258                 }
 259
 260                 $a = DI::app();
 261                 $ch = curl_init($url);
 262
 263                 if (($redirects > 8) || (!$ch)) {
 264                         return CurlResult::createErrorCurl($url);
 265                 }
 266
 267                 Logger::log('post_url: start ' . $url, Logger::DATA);
 268
 269                 curl_setopt($ch, CURLOPT_HEADER, true);
 270                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 271                 curl_setopt($ch, CURLOPT_POST, 1);
 272                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 273                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 274
 275                 if (DI::config()->get('system', 'ipv4_resolve', false)) {
 276                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 277                 }
 278
 279                 if (intval($timeout)) {
 280                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 281                 } else {
 282                         $curl_time = DI::config()->get('system', 'curl_timeout', 60);
 283                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 284                 }
 285
 286                 if (!empty($headers)) {
 287                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 288                 }
 289
 290                 $check_cert = DI::config()->get('system', 'verifyssl');
 291                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 292
 293                 if ($check_cert) {
 294                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 295                 }
 296
 297                 $proxy = DI::config()->get('system', 'proxy');
 298
 299                 if (strlen($proxy)) {
 300                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 301                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 302                         $proxyuser = DI::config()->get('system', 'proxyuser');
 303                         if (strlen($proxyuser)) {
 304                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 305                         }
 306                 }
 307
 308                 // don't let curl abort the entire application
 309                 // if it throws any errors.
 310
 311                 $s = @curl_exec($ch);
 312
 313                 $curl_info = curl_getinfo($ch);
 314
 315                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 316
 317                 if ($curlResponse->isRedirectUrl()) {
 318                         $redirects++;
 319                         Logger::log('post_url: redirect ' . $url . ' to ' . $curlResponse->getRedirectUrl());
 320                         curl_close($ch);
 321                         return self::post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 322                 }
 323
 324                 curl_close($ch);
 325
 326                 DI::profiler()->saveTimestamp($stamp1, 'network', System::callstack());
 327
 328                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 329                 if ($curlResponse->getReturnCode() == 417) {
 330                         $redirects++;
 331
 332                         if (empty($headers)) {
 333                                 $headers = ['Expect:'];
 334                         } else {
 335                                 if (!in_array('Expect:', $headers)) {
 336                                         array_push($headers, 'Expect:');
 337                                 }
 338                         }
 339                         Logger::info('Server responds with 417, applying workaround', ['url' => $url]);
 340                         return self::post($url, $params, $headers, $redirects, $timeout);
 341                 }
 342
 343                 Logger::log('post_url: end ' . $url, Logger::DATA);
 344
 345                 return $curlResponse;
 346         }
 347
 348         /**
 349          * Return raw post data from a post request
 350          *
 351          * @return string post data
 352          */
 353         public static function postdata()
 354         {
 355                 return file_get_contents('php://input');
 356         }
 357
 358         /**
 359          * Check URL to see if it's real
 360          *
 361          * Take a URL from the wild, prepend http:// if necessary
 362          * and check DNS to see if it's real (or check if is a valid IP address)
 363          *
 364          * @param string $url The URL to be validated
 365          * @return string|boolean The actual working URL, false else
 366          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 367          */
 368         public static function isUrlValid(string $url)
 369         {
 370                 if (DI::config()->get('system', 'disable_url_validation')) {
 371                         return $url;
 372                 }
 373
 374                 // no naked subdomains (allow localhost for tests)
 375                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 376                         return false;
 377                 }
 378
 379                 if (substr($url, 0, 4) != 'http') {
 380                         $url = 'http://' . $url;
 381                 }
 382
 383                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 384                 $h = @parse_url($url);
 385
 386                 if (!empty($h['host']) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP))) {
 387                         return $url;
 388                 }
 389
 390                 return false;
 391         }
 392
 393         /**
 394          * Checks that email is an actual resolvable internet address
 395          *
 396          * @param string $addr The email address
 397          * @return boolean True if it's a valid email address, false if it's not
 398          */
 399         public static function isEmailDomainValid(string $addr)
 400         {
 401                 if (DI::config()->get('system', 'disable_email_validation')) {
 402                         return true;
 403                 }
 404
 405                 if (! strpos($addr, '@')) {
 406                         return false;
 407                 }
 408
 409                 $h = substr($addr, strpos($addr, '@') + 1);
 410
 411                 // Concerning the @ see here: https://stackoverflow.com/questions/36280957/dns-get-record-a-temporary-server-error-occurred
 412                 if ($h && (@dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP))) {
 413                         return true;
 414                 }
 415                 if ($h && @dns_get_record($h, DNS_CNAME + DNS_MX)) {
 416                         return true;
 417                 }
 418                 return false;
 419         }
 420
 421         /**
 422          * Check if URL is allowed
 423          *
 424          * Check $url against our list of allowed sites,
 425          * wildcards allowed. If allowed_sites is unset return true;
 426          *
 427          * @param string $url URL which get tested
 428          * @return boolean True if url is allowed otherwise return false
 429          */
 430         public static function isUrlAllowed(string $url)
 431         {
 432                 $h = @parse_url($url);
 433
 434                 if (! $h) {
 435                         return false;
 436                 }
 437
 438                 $str_allowed = DI::config()->get('system', 'allowed_sites');
 439                 if (! $str_allowed) {
 440                         return true;
 441                 }
 442
 443                 $found = false;
 444
 445                 $host = strtolower($h['host']);
 446
 447                 // always allow our own site
 448                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 449                         return true;
 450                 }
 451
 452                 $fnmatch = function_exists('fnmatch');
 453                 $allowed = explode(',', $str_allowed);
 454
 455                 if (count($allowed)) {
 456                         foreach ($allowed as $a) {
 457                                 $pat = strtolower(trim($a));
 458                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 459                                         $found = true;
 460                                         break;
 461                                 }
 462                         }
 463                 }
 464                 return $found;
 465         }
 466
 467         /**
 468          * Checks if the provided url domain is on the domain blocklist.
 469          * Returns true if it is or malformed URL, false if not.
 470          *
 471          * @param string $url The url to check the domain from
 472          *
 473          * @return boolean
 474          */
 475         public static function isUrlBlocked(string $url)
 476         {
 477                 $host = @parse_url($url, PHP_URL_HOST);
 478                 if (!$host) {
 479                         return false;
 480                 }
 481
 482                 $domain_blocklist = DI::config()->get('system', 'blocklist', []);
 483                 if (!$domain_blocklist) {
 484                         return false;
 485                 }
 486
 487                 foreach ($domain_blocklist as $domain_block) {
 488                         if (fnmatch(strtolower($domain_block['domain']), strtolower($host))) {
 489                                 return true;
 490                         }
 491                 }
 492
 493                 return false;
 494         }
 495
 496         /**
 497          * Check if email address is allowed to register here.
 498          *
 499          * Compare against our list (wildcards allowed).
 500          *
 501          * @param  string $email email address
 502          * @return boolean False if not allowed, true if allowed
 503          *                       or if allowed list is not configured
 504          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 505          */
 506         public static function isEmailDomainAllowed(string $email)
 507         {
 508                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 509                 if (!$domain) {
 510                         return false;
 511                 }
 512
 513                 $str_allowed = DI::config()->get('system', 'allowed_email', '');
 514                 if (empty($str_allowed)) {
 515                         return true;
 516                 }
 517
 518                 $allowed = explode(',', $str_allowed);
 519
 520                 return self::isDomainAllowed($domain, $allowed);
 521         }
 522
 523         /**
 524          * Checks for the existence of a domain in a domain list
 525          *
 526          * @param string $domain
 527          * @param array  $domain_list
 528          * @return boolean
 529          */
 530         public static function isDomainAllowed(string $domain, array $domain_list)
 531         {
 532                 $found = false;
 533
 534                 foreach ($domain_list as $item) {
 535                         $pat = strtolower(trim($item));
 536                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 537                                 $found = true;
 538                                 break;
 539                         }
 540                 }
 541
 542                 return $found;
 543         }
 544
 545         public static function lookupAvatarByEmail(string $email)
 546         {
 547                 $avatar['size'] = 300;
 548                 $avatar['email'] = $email;
 549                 $avatar['url'] = '';
 550                 $avatar['success'] = false;
 551
 552                 Hook::callAll('avatar_lookup', $avatar);
 553
 554                 if (! $avatar['success']) {
 555                         $avatar['url'] = DI::baseUrl() . '/images/person-300.jpg';
 556                 }
 557
 558                 Logger::log('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], Logger::DEBUG);
 559                 return $avatar['url'];
 560         }
 561
 562         /**
 563          * Remove Google Analytics and other tracking platforms params from URL
 564          *
 565          * @param string $url Any user-submitted URL that may contain tracking params
 566          * @return string The same URL stripped of tracking parameters
 567          */
 568         public static function stripTrackingQueryParams(string $url)
 569         {
 570                 $urldata = parse_url($url);
 571                 if (!empty($urldata["query"])) {
 572                         $query = $urldata["query"];
 573                         parse_str($query, $querydata);
 574
 575                         if (is_array($querydata)) {
 576                                 foreach ($querydata as $param => $value) {
 577                                         if (in_array(
 578                                                 $param,
 579                                                 [
 580                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 581                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 582                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 583                                                         "awesm", "wtrid",
 584                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 585                                                 )
 586                                         ) {
 587                                                 $pair = $param . "=" . urlencode($value);
 588                                                 $url = str_replace($pair, "", $url);
 589
 590                                                 // Second try: if the url isn't encoded completely
 591                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 592                                                 $url = str_replace($pair, "", $url);
 593
 594                                                 // Third try: Maybey the url isn't encoded at all
 595                                                 $pair = $param . "=" . $value;
 596                                                 $url = str_replace($pair, "", $url);
 597
 598                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 599                                         }
 600                                 }
 601                         }
 602
 603                         if (substr($url, -1, 1) == "?") {
 604                                 $url = substr($url, 0, -1);
 605                         }
 606                 }
 607
 608                 return $url;
 609         }
 610
 611         /**
 612          * Returns the original URL of the provided URL
 613          *
 614          * This function strips tracking query params and follows redirections, either
 615          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 616          *
 617          * @todo  Remove the $fetchbody parameter that generates an extraneous HEAD request
 618          *
 619          * @see   ParseUrl::getSiteinfo
 620          *
 621          * @param string $url       A user-submitted URL
 622          * @param int    $depth     The current redirection recursion level (internal)
 623          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 624          * @return string A canonical URL
 625          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 626          */
 627         public static function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 628         {
 629                 $a = DI::app();
 630
 631                 $url = self::stripTrackingQueryParams($url);
 632
 633                 if ($depth > 10) {
 634                         return $url;
 635                 }
 636
 637                 $url = trim($url, "'");
 638
 639                 $stamp1 = microtime(true);
 640
 641                 $ch = curl_init();
 642                 curl_setopt($ch, CURLOPT_URL, $url);
 643                 curl_setopt($ch, CURLOPT_HEADER, 1);
 644                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 645                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 646                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 647                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 648
 649                 curl_exec($ch);
 650                 $curl_info = @curl_getinfo($ch);
 651                 $http_code = $curl_info['http_code'];
 652                 curl_close($ch);
 653
 654                 DI::profiler()->saveTimestamp($stamp1, "network", System::callstack());
 655
 656                 if ($http_code == 0) {
 657                         return $url;
 658                 }
 659
 660                 if (in_array($http_code, ['301', '302'])) {
 661                         if (!empty($curl_info['redirect_url'])) {
 662                                 return self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 663                         } elseif (!empty($curl_info['location'])) {
 664                                 return self::finalUrl($curl_info['location'], ++$depth, $fetchbody);
 665                         }
 666                 }
 667
 668                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 669                 if (!$fetchbody) {
 670                         return(self::finalUrl($url, ++$depth, true));
 671                 }
 672
 673                 // if the file is too large then exit
 674                 if ($curl_info["download_content_length"] > 1000000) {
 675                         return $url;
 676                 }
 677
 678                 // if it isn't a HTML file then exit
 679                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 680                         return $url;
 681                 }
 682
 683                 $stamp1 = microtime(true);
 684
 685                 $ch = curl_init();
 686                 curl_setopt($ch, CURLOPT_URL, $url);
 687                 curl_setopt($ch, CURLOPT_HEADER, 0);
 688                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 689                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 690                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 691                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 692
 693                 $body = curl_exec($ch);
 694                 curl_close($ch);
 695
 696                 DI::profiler()->saveTimestamp($stamp1, "network", System::callstack());
 697
 698                 if (trim($body) == "") {
 699                         return $url;
 700                 }
 701
 702                 // Check for redirect in meta elements
 703                 $doc = new DOMDocument();
 704                 @$doc->loadHTML($body);
 705
 706                 $xpath = new DomXPath($doc);
 707
 708                 $list = $xpath->query("//meta[@content]");
 709                 foreach ($list as $node) {
 710                         $attr = [];
 711                         if ($node->attributes->length) {
 712                                 foreach ($node->attributes as $attribute) {
 713                                         $attr[$attribute->name] = $attribute->value;
 714                                 }
 715                         }
 716
 717                         if (@$attr["http-equiv"] == 'refresh') {
 718                                 $path = $attr["content"];
 719                                 $pathinfo = explode(";", $path);
 720                                 foreach ($pathinfo as $value) {
 721                                         if (substr(strtolower($value), 0, 4) == "url=") {
 722                                                 return self::finalUrl(substr($value, 4), ++$depth);
 723                                         }
 724                                 }
 725                         }
 726                 }
 727
 728                 return $url;
 729         }
 730
 731         /**
 732          * Find the matching part between two url
 733          *
 734          * @param string $url1
 735          * @param string $url2
 736          * @return string The matching part
 737          */
 738         public static function getUrlMatch(string $url1, string $url2)
 739         {
 740                 if (($url1 == "") || ($url2 == "")) {
 741                         return "";
 742                 }
 743
 744                 $url1 = Strings::normaliseLink($url1);
 745                 $url2 = Strings::normaliseLink($url2);
 746
 747                 $parts1 = parse_url($url1);
 748                 $parts2 = parse_url($url2);
 749
 750                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 751                         return "";
 752                 }
 753
 754                 if (empty($parts1["scheme"])) {
 755                         $parts1["scheme"] = '';
 756                 }
 757                 if (empty($parts2["scheme"])) {
 758                         $parts2["scheme"] = '';
 759                 }
 760
 761                 if ($parts1["scheme"] != $parts2["scheme"]) {
 762                         return "";
 763                 }
 764
 765                 if (empty($parts1["host"])) {
 766                         $parts1["host"] = '';
 767                 }
 768                 if (empty($parts2["host"])) {
 769                         $parts2["host"] = '';
 770                 }
 771
 772                 if ($parts1["host"] != $parts2["host"]) {
 773                         return "";
 774                 }
 775
 776                 if (empty($parts1["port"])) {
 777                         $parts1["port"] = '';
 778                 }
 779                 if (empty($parts2["port"])) {
 780                         $parts2["port"] = '';
 781                 }
 782
 783                 if ($parts1["port"] != $parts2["port"]) {
 784                         return "";
 785                 }
 786
 787                 $match = $parts1["scheme"]."://".$parts1["host"];
 788
 789                 if ($parts1["port"]) {
 790                         $match .= ":".$parts1["port"];
 791                 }
 792
 793                 if (empty($parts1["path"])) {
 794                         $parts1["path"] = '';
 795                 }
 796                 if (empty($parts2["path"])) {
 797                         $parts2["path"] = '';
 798                 }
 799
 800                 $pathparts1 = explode("/", $parts1["path"]);
 801                 $pathparts2 = explode("/", $parts2["path"]);
 802
 803                 $i = 0;
 804                 $path = "";
 805                 do {
 806                         $path1 = $pathparts1[$i] ?? '';
 807                         $path2 = $pathparts2[$i] ?? '';
 808
 809                         if ($path1 == $path2) {
 810                                 $path .= $path1."/";
 811                         }
 812                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 813
 814                 $match .= $path;
 815
 816                 return Strings::normaliseLink($match);
 817         }
 818
 819         /**
 820          * Glue url parts together
 821          *
 822          * @param array $parsed URL parts
 823          *
 824          * @return string The glued URL
 825          */
 826         public static function unparseURL(array $parsed)
 827         {
 828                 $get = function ($key) use ($parsed) {
 829                         return isset($parsed[$key]) ? $parsed[$key] : null;
 830                 };
 831
 832                 $pass      = $get('pass');
 833                 $user      = $get('user');
 834                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 835                 $port      = $get('port');
 836                 $scheme    = $get('scheme');
 837                 $query     = $get('query');
 838                 $fragment  = $get('fragment');
 839                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 840                                                 $get('host') .
 841                                                 ($port ? ":$port" : '');
 842
 843                 return  (strlen($scheme) ? $scheme.":" : '') .
 844                         (strlen($authority) ? "//".$authority : '') .
 845                         $get('path') .
 846                         (strlen($query) ? "?".$query : '') .
 847                         (strlen($fragment) ? "#".$fragment : '');
 848         }
 849
 850
 851         /**
 852          * Switch the scheme of an url between http and https
 853          *
 854          * @param string $url URL
 855          *
 856          * @return string switched URL
 857          */
 858         public static function switchScheme(string $url)
 859         {
 860                 $scheme = parse_url($url, PHP_URL_SCHEME);
 861                 if (empty($scheme)) {
 862                         return $url;
 863                 }
 864
 865                 if ($scheme === 'http') {
 866                         $url = str_replace('http://', 'https://', $url);
 867                 } elseif ($scheme === 'https') {
 868                         $url = str_replace('https://', 'http://', $url);
 869                 }
 870
 871                 return $url;
 872         }
 873
 874         /**
 875          * Adds query string parameters to the provided URI. Replace the value of existing keys.
 876          *
 877          * @param string $path
 878          * @param array  $additionalParams Associative array of parameters
 879          * @return string
 880          */
 881         public static function appendQueryParam(string $path, array $additionalParams)
 882         {
 883                 $parsed = parse_url($path);
 884
 885                 $params = [];
 886                 if (!empty($parsed['query'])) {
 887                         parse_str($parsed['query'], $params);
 888                 }
 889
 890                 $params = array_merge($params, $additionalParams);
 891
 892                 $parsed['query'] = http_build_query($params);
 893
 894                 return self::unparseURL($parsed);
 895         }
 896 }