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