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\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          * @brief Curl wrapper
  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     $redirects      The recursion counter for internal use - default 0
  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          *
  33          * @return string The fetched content
  34          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  35          */
  36         public static function fetchUrl(string $url, bool $binary = false, int &$redirects = 0, int $timeout = 0, string $accept_content = null, string $cookiejar = '')
  37         {
  38                 $ret = self::fetchUrlFull($url, $binary, $redirects, $timeout, $accept_content, $cookiejar);
  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          * @brief Curl wrapper with array of return values.
  50          * @param string  $url            URL to fetch
  51          * @param bool    $binary         default false
  52          *                                TRUE if asked to return binary results (file download)
  53          * @param int     $redirects      The recursion counter for internal use - default 0
  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          *
  58          * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
  59          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  60          */
  61         public static function fetchUrlFull(string $url, bool $binary = false, int &$redirects = 0, int $timeout = 0, string $accept_content = null, string $cookiejar = '')
  62         {
  63                 return self::curl(
  64                         $url,
  65                         $binary,
  66                         $redirects,
  67                         ['timeout'=>$timeout,
  68                         'accept_content'=>$accept_content,
  69                         'cookiejar'=>$cookiejar
  70                         ]
  71                 );
  72         }
  73
  74         /**
  75          * @brief 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 int     $redirects The recursion counter for internal use - default 0
  81          * @param array   $opts      (optional parameters) assoziative array with:
  82          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  83          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  84          *                           'http_auth' => username:password
  85          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  86          *                           'nobody' => only return the header
  87          *                           'cookiejar' => path to cookie jar file
  88          *                           'header' => header array
  89          *
  90          * @return CurlResult
  91          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  92          */
  93         public static function curl(string $url, bool $binary = false, int &$redirects = 0, array $opts = [])
  94         {
  95                 $stamp1 = microtime(true);
  96
  97                 $a = \get_app();
  98
  99                 if (strlen($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                 $path_parts = explode('/', defaults($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, $redirects, $opts);
 231                 }
 232
 233                 @curl_close($ch);
 234
 235                 $a->getProfiler()->saveTimestamp($stamp1, 'network', System::callstack());
 236
 237                 return $curlResponse;
 238         }
 239
 240         /**
 241          * @brief Send POST request to $url
 242          *
 243          * @param string  $url       URL to post
 244          * @param mixed   $params    array of POST variables
 245          * @param string  $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, string $headers = null, int &$redirects = 0, int $timeout = 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 = \get_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($curl_time));
 285                 }
 286
 287                 if (defined('LIGHTTPD')) {
 288                         if (!is_array($headers)) {
 289                                 $headers = ['Expect:'];
 290                         } else {
 291                                 if (!in_array('Expect:', $headers)) {
 292                                         array_push($headers, 'Expect:');
 293                                 }
 294                         }
 295                 }
 296
 297                 if ($headers) {
 298                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 299                 }
 300
 301                 $check_cert = Config::get('system', 'verifyssl');
 302                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 303
 304                 if ($check_cert) {
 305                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 306                 }
 307
 308                 $proxy = Config::get('system', 'proxy');
 309
 310                 if (strlen($proxy)) {
 311                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 312                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 313                         $proxyuser = Config::get('system', 'proxyuser');
 314                         if (strlen($proxyuser)) {
 315                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 316                         }
 317                 }
 318
 319                 // don't let curl abort the entire application
 320                 // if it throws any errors.
 321
 322                 $s = @curl_exec($ch);
 323
 324                 $curl_info = curl_getinfo($ch);
 325
 326                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 327
 328                 if ($curlResponse->isRedirectUrl()) {
 329                         $redirects++;
 330                         Logger::log('post_url: redirect ' . $url . ' to ' . $curlResponse->getRedirectUrl());
 331                         curl_close($ch);
 332                         return self::post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 333                 }
 334
 335                 curl_close($ch);
 336
 337                 $a->getProfiler()->saveTimestamp($stamp1, 'network', System::callstack());
 338
 339                 Logger::log('post_url: end ' . $url, Logger::DATA);
 340
 341                 return $curlResponse;
 342         }
 343
 344         /**
 345          * @brief Check URL to see if it's real
 346          *
 347          * Take a URL from the wild, prepend http:// if necessary
 348          * and check DNS to see if it's real (or check if is a valid IP address)
 349          *
 350          * @param string $url The URL to be validated
 351          * @return string|boolean The actual working URL, false else
 352          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 353          */
 354         public static function isUrlValid(string $url)
 355         {
 356                 if (Config::get('system', 'disable_url_validation')) {
 357                         return $url;
 358                 }
 359
 360                 // no naked subdomains (allow localhost for tests)
 361                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 362                         return false;
 363                 }
 364
 365                 if (substr($url, 0, 4) != 'http') {
 366                         $url = 'http://' . $url;
 367                 }
 368
 369                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 370                 $h = @parse_url($url);
 371
 372                 if (!empty($h['host']) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 373                         return $url;
 374                 }
 375
 376                 return false;
 377         }
 378
 379         /**
 380          * @brief Checks that email is an actual resolvable internet address
 381          *
 382          * @param string $addr The email address
 383          * @return boolean True if it's a valid email address, false if it's not
 384          */
 385         public static function isEmailDomainValid(string $addr)
 386         {
 387                 if (Config::get('system', 'disable_email_validation')) {
 388                         return true;
 389                 }
 390
 391                 if (! strpos($addr, '@')) {
 392                         return false;
 393                 }
 394
 395                 $h = substr($addr, strpos($addr, '@') + 1);
 396
 397                 // Concerning the @ see here: https://stackoverflow.com/questions/36280957/dns-get-record-a-temporary-server-error-occurred
 398                 if ($h && (@dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 399                         return true;
 400                 }
 401                 if ($h && @dns_get_record($h, DNS_CNAME + DNS_MX)) {
 402                         return true;
 403                 }
 404                 return false;
 405         }
 406
 407         /**
 408          * @brief Check if URL is allowed
 409          *
 410          * Check $url against our list of allowed sites,
 411          * wildcards allowed. If allowed_sites is unset return true;
 412          *
 413          * @param string $url URL which get tested
 414          * @return boolean True if url is allowed otherwise return false
 415          */
 416         public static function isUrlAllowed(string $url)
 417         {
 418                 $h = @parse_url($url);
 419
 420                 if (! $h) {
 421                         return false;
 422                 }
 423
 424                 $str_allowed = Config::get('system', 'allowed_sites');
 425                 if (! $str_allowed) {
 426                         return true;
 427                 }
 428
 429                 $found = false;
 430
 431                 $host = strtolower($h['host']);
 432
 433                 // always allow our own site
 434                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 435                         return true;
 436                 }
 437
 438                 $fnmatch = function_exists('fnmatch');
 439                 $allowed = explode(',', $str_allowed);
 440
 441                 if (count($allowed)) {
 442                         foreach ($allowed as $a) {
 443                                 $pat = strtolower(trim($a));
 444                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 445                                         $found = true;
 446                                         break;
 447                                 }
 448                         }
 449                 }
 450                 return $found;
 451         }
 452
 453         /**
 454          * Checks if the provided url domain is on the domain blocklist.
 455          * Returns true if it is or malformed URL, false if not.
 456          *
 457          * @param string $url The url to check the domain from
 458          *
 459          * @return boolean
 460          */
 461         public static function isUrlBlocked(string $url)
 462         {
 463                 $host = @parse_url($url, PHP_URL_HOST);
 464                 if (!$host) {
 465                         return false;
 466                 }
 467
 468                 $domain_blocklist = Config::get('system', 'blocklist', []);
 469                 if (!$domain_blocklist) {
 470                         return false;
 471                 }
 472
 473                 foreach ($domain_blocklist as $domain_block) {
 474                         if (strcasecmp($domain_block['domain'], $host) === 0) {
 475                                 return true;
 476                         }
 477                 }
 478
 479                 return false;
 480         }
 481
 482         /**
 483          * @brief Check if email address is allowed to register here.
 484          *
 485          * Compare against our list (wildcards allowed).
 486          *
 487          * @param  string $email email address
 488          * @return boolean False if not allowed, true if allowed
 489          *                       or if allowed list is not configured
 490          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 491          */
 492         public static function isEmailDomainAllowed(string $email)
 493         {
 494                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 495                 if (!$domain) {
 496                         return false;
 497                 }
 498
 499                 $str_allowed = Config::get('system', 'allowed_email', '');
 500                 if (empty($str_allowed)) {
 501                         return true;
 502                 }
 503
 504                 $allowed = explode(',', $str_allowed);
 505
 506                 return self::isDomainAllowed($domain, $allowed);
 507         }
 508
 509         /**
 510          * Checks for the existence of a domain in a domain list
 511          *
 512          * @brief Checks for the existence of a domain in a domain list
 513          * @param string $domain
 514          * @param array  $domain_list
 515          * @return boolean
 516          */
 517         public static function isDomainAllowed(string $domain, array $domain_list)
 518         {
 519                 $found = false;
 520
 521                 foreach ($domain_list as $item) {
 522                         $pat = strtolower(trim($item));
 523                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 524                                 $found = true;
 525                                 break;
 526                         }
 527                 }
 528
 529                 return $found;
 530         }
 531
 532         public static function lookupAvatarByEmail(string $email)
 533         {
 534                 $avatar['size'] = 300;
 535                 $avatar['email'] = $email;
 536                 $avatar['url'] = '';
 537                 $avatar['success'] = false;
 538
 539                 Hook::callAll('avatar_lookup', $avatar);
 540
 541                 if (! $avatar['success']) {
 542                         $avatar['url'] = System::baseUrl() . '/images/person-300.jpg';
 543                 }
 544
 545                 Logger::log('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], Logger::DEBUG);
 546                 return $avatar['url'];
 547         }
 548
 549         /**
 550          * @brief Remove Google Analytics and other tracking platforms params from URL
 551          *
 552          * @param string $url Any user-submitted URL that may contain tracking params
 553          * @return string The same URL stripped of tracking parameters
 554          */
 555         public static function stripTrackingQueryParams(string $url)
 556         {
 557                 $urldata = parse_url($url);
 558                 if (!empty($urldata["query"])) {
 559                         $query = $urldata["query"];
 560                         parse_str($query, $querydata);
 561
 562                         if (is_array($querydata)) {
 563                                 foreach ($querydata as $param => $value) {
 564                                         if (in_array(
 565                                                 $param,
 566                                                 [
 567                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 568                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 569                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 570                                                         "awesm", "wtrid",
 571                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 572                                                 )
 573                                         ) {
 574                                                 $pair = $param . "=" . urlencode($value);
 575                                                 $url = str_replace($pair, "", $url);
 576
 577                                                 // Second try: if the url isn't encoded completely
 578                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 579                                                 $url = str_replace($pair, "", $url);
 580
 581                                                 // Third try: Maybey the url isn't encoded at all
 582                                                 $pair = $param . "=" . $value;
 583                                                 $url = str_replace($pair, "", $url);
 584
 585                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 586                                         }
 587                                 }
 588                         }
 589
 590                         if (substr($url, -1, 1) == "?") {
 591                                 $url = substr($url, 0, -1);
 592                         }
 593                 }
 594
 595                 return $url;
 596         }
 597
 598         /**
 599          * @brief Returns the original URL of the provided URL
 600          *
 601          * This function strips tracking query params and follows redirections, either
 602          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 603          *
 604          * @todo  Remove the $fetchbody parameter that generates an extraneous HEAD request
 605          *
 606          * @see   ParseUrl::getSiteinfo
 607          *
 608          * @param string $url       A user-submitted URL
 609          * @param int    $depth     The current redirection recursion level (internal)
 610          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 611          * @return string A canonical URL
 612          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 613          */
 614         public static function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 615         {
 616                 $a = \get_app();
 617
 618                 $url = self::stripTrackingQueryParams($url);
 619
 620                 if ($depth > 10) {
 621                         return $url;
 622                 }
 623
 624                 $url = trim($url, "'");
 625
 626                 $stamp1 = microtime(true);
 627
 628                 $ch = curl_init();
 629                 curl_setopt($ch, CURLOPT_URL, $url);
 630                 curl_setopt($ch, CURLOPT_HEADER, 1);
 631                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 632                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 633                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 634                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 635
 636                 curl_exec($ch);
 637                 $curl_info = @curl_getinfo($ch);
 638                 $http_code = $curl_info['http_code'];
 639                 curl_close($ch);
 640
 641                 $a->getProfiler()->saveTimestamp($stamp1, "network", System::callstack());
 642
 643                 if ($http_code == 0) {
 644                         return $url;
 645                 }
 646
 647                 if (in_array($http_code, ['301', '302'])) {
 648                         if (!empty($curl_info['redirect_url'])) {
 649                                 return self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 650                         } elseif (!empty($curl_info['location'])) {
 651                                 return self::finalUrl($curl_info['location'], ++$depth, $fetchbody);
 652                         }
 653                 }
 654
 655                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 656                 if (!$fetchbody) {
 657                         return(self::finalUrl($url, ++$depth, true));
 658                 }
 659
 660                 // if the file is too large then exit
 661                 if ($curl_info["download_content_length"] > 1000000) {
 662                         return $url;
 663                 }
 664
 665                 // if it isn't a HTML file then exit
 666                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 667                         return $url;
 668                 }
 669
 670                 $stamp1 = microtime(true);
 671
 672                 $ch = curl_init();
 673                 curl_setopt($ch, CURLOPT_URL, $url);
 674                 curl_setopt($ch, CURLOPT_HEADER, 0);
 675                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 676                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 677                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 678                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 679
 680                 $body = curl_exec($ch);
 681                 curl_close($ch);
 682
 683                 $a->getProfiler()->saveTimestamp($stamp1, "network", System::callstack());
 684
 685                 if (trim($body) == "") {
 686                         return $url;
 687                 }
 688
 689                 // Check for redirect in meta elements
 690                 $doc = new DOMDocument();
 691                 @$doc->loadHTML($body);
 692
 693                 $xpath = new DomXPath($doc);
 694
 695                 $list = $xpath->query("//meta[@content]");
 696                 foreach ($list as $node) {
 697                         $attr = [];
 698                         if ($node->attributes->length) {
 699                                 foreach ($node->attributes as $attribute) {
 700                                         $attr[$attribute->name] = $attribute->value;
 701                                 }
 702                         }
 703
 704                         if (@$attr["http-equiv"] == 'refresh') {
 705                                 $path = $attr["content"];
 706                                 $pathinfo = explode(";", $path);
 707                                 foreach ($pathinfo as $value) {
 708                                         if (substr(strtolower($value), 0, 4) == "url=") {
 709                                                 return self::finalUrl(substr($value, 4), ++$depth);
 710                                         }
 711                                 }
 712                         }
 713                 }
 714
 715                 return $url;
 716         }
 717
 718         /**
 719          * @brief Find the matching part between two url
 720          *
 721          * @param string $url1
 722          * @param string $url2
 723          * @return string The matching part
 724          */
 725         public static function getUrlMatch(string $url1, string $url2)
 726         {
 727                 if (($url1 == "") || ($url2 == "")) {
 728                         return "";
 729                 }
 730
 731                 $url1 = Strings::normaliseLink($url1);
 732                 $url2 = Strings::normaliseLink($url2);
 733
 734                 $parts1 = parse_url($url1);
 735                 $parts2 = parse_url($url2);
 736
 737                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 738                         return "";
 739                 }
 740
 741                 if (empty($parts1["scheme"])) {
 742                         $parts1["scheme"] = '';
 743                 }
 744                 if (empty($parts2["scheme"])) {
 745                         $parts2["scheme"] = '';
 746                 }
 747
 748                 if ($parts1["scheme"] != $parts2["scheme"]) {
 749                         return "";
 750                 }
 751
 752                 if (empty($parts1["host"])) {
 753                         $parts1["host"] = '';
 754                 }
 755                 if (empty($parts2["host"])) {
 756                         $parts2["host"] = '';
 757                 }
 758
 759                 if ($parts1["host"] != $parts2["host"]) {
 760                         return "";
 761                 }
 762
 763                 if (empty($parts1["port"])) {
 764                         $parts1["port"] = '';
 765                 }
 766                 if (empty($parts2["port"])) {
 767                         $parts2["port"] = '';
 768                 }
 769
 770                 if ($parts1["port"] != $parts2["port"]) {
 771                         return "";
 772                 }
 773
 774                 $match = $parts1["scheme"]."://".$parts1["host"];
 775
 776                 if ($parts1["port"]) {
 777                         $match .= ":".$parts1["port"];
 778                 }
 779
 780                 if (empty($parts1["path"])) {
 781                         $parts1["path"] = '';
 782                 }
 783                 if (empty($parts2["path"])) {
 784                         $parts2["path"] = '';
 785                 }
 786
 787                 $pathparts1 = explode("/", $parts1["path"]);
 788                 $pathparts2 = explode("/", $parts2["path"]);
 789
 790                 $i = 0;
 791                 $path = "";
 792                 do {
 793                         $path1 = defaults($pathparts1, $i, '');
 794                         $path2 = defaults($pathparts2, $i, '');
 795
 796                         if ($path1 == $path2) {
 797                                 $path .= $path1."/";
 798                         }
 799                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 800
 801                 $match .= $path;
 802
 803                 return Strings::normaliseLink($match);
 804         }
 805
 806         /**
 807          * @brief Glue url parts together
 808          *
 809          * @param array $parsed URL parts
 810          *
 811          * @return string The glued URL
 812          */
 813         public static function unparseURL(array $parsed)
 814         {
 815                 $get = function ($key) use ($parsed) {
 816                         return isset($parsed[$key]) ? $parsed[$key] : null;
 817                 };
 818
 819                 $pass      = $get('pass');
 820                 $user      = $get('user');
 821                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 822                 $port      = $get('port');
 823                 $scheme    = $get('scheme');
 824                 $query     = $get('query');
 825                 $fragment  = $get('fragment');
 826                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 827                                                 $get('host') .
 828                                                 ($port ? ":$port" : '');
 829
 830                 return  (strlen($scheme) ? $scheme.":" : '') .
 831                         (strlen($authority) ? "//".$authority : '') .
 832                         $get('path') .
 833                         (strlen($query) ? "?".$query : '') .
 834                         (strlen($fragment) ? "#".$fragment : '');
 835         }
 836
 837
 838         /**
 839          * Switch the scheme of an url between http and https
 840          *
 841          * @param string $url URL
 842          *
 843          * @return string switched URL
 844          */
 845         public static function switchScheme(string $url)
 846         {
 847                 $scheme = parse_url($url, PHP_URL_SCHEME);
 848                 if (empty($scheme)) {
 849                         return $url;
 850                 }
 851
 852                 if ($scheme === 'http') {
 853                         $url = str_replace('http://', 'https://', $url);
 854                 } elseif ($scheme === 'https') {
 855                         $url = str_replace('https://', 'http://', $url);
 856                 }
 857
 858                 return $url;
 859         }
 860 }