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