src/Util/Network.php

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