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