src/Util/Network.php

   1 <?php
   2 /**
   3  * @file src/Util/Network.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use Friendica\Core\Addon;
   8 use Friendica\Core\Logger;
   9 use Friendica\Core\System;
  10 use Friendica\Core\Config;
  11 use Friendica\Network\CurlResult;
  12 use Friendica\Util\Strings;
  13 use DOMDocument;
  14 use DomXPath;
  15
  16 class Network
  17 {
  18         /**
  19          * Curl wrapper
  20          *
  21          * If binary flag is true, return binary results.
  22          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
  23          * to preserve cookies from one request to the next.
  24          *
  25          * @brief Curl wrapper
  26          * @param string  $url            URL to fetch
  27          * @param boolean $binary         default false
  28          *                                TRUE if asked to return binary results (file download)
  29          * @param integer $redirects      The recursion counter for internal use - default 0
  30          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  31          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  32          * @param string  $cookiejar      Path to cookie jar file
  33          *
  34          * @return string The fetched content
  35          */
  36         public static function fetchUrl($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  $cookiejar      Path to cookie jar file
  57          *
  58          * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
  59          */
  60         public static function fetchUrlFull($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = '')
  61         {
  62                 return self::curl(
  63                         $url,
  64                         $binary,
  65                         $redirects,
  66                         ['timeout'=>$timeout,
  67                         'accept_content'=>$accept_content,
  68                         'cookiejar'=>$cookiejar
  69                         ]
  70                 );
  71         }
  72
  73         /**
  74          * @brief fetches an URL.
  75          *
  76          * @param string  $url       URL to fetch
  77          * @param boolean $binary    default false
  78          *                           TRUE if asked to return binary results (file download)
  79          * @param int     $redirects The recursion counter for internal use - default 0
  80          * @param array   $opts      (optional parameters) assoziative array with:
  81          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  82          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  83          *                           'http_auth' => username:password
  84          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  85          *                           'nobody' => only return the header
  86          *                           'cookiejar' => path to cookie jar file
  87          *                           'header' => header array
  88          *
  89          * @return CurlResult
  90          */
  91         public static function curl($url, $binary = false, &$redirects = 0, $opts = [])
  92         {
  93                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
  94
  95                 $stamp1 = microtime(true);
  96
  97                 $a = get_app();
  98
  99                 $parts = parse_url($url);
 100                 $path_parts = explode('/', defaults($parts, 'path', ''));
 101                 foreach ($path_parts as $part) {
 102                         if (strlen($part) <> mb_strlen($part)) {
 103                                 $parts2[] = rawurlencode($part);
 104                         } else {
 105                                 $parts2[] = $part;
 106                         }
 107                 }
 108                 $parts['path'] = implode('/', $parts2);
 109                 $url = self::unparseURL($parts);
 110
 111                 if (self::isUrlBlocked($url)) {
 112                         Logger::log('domain of ' . $url . ' is blocked', Logger::DATA);
 113                         return CurlResult::createErrorCurl($url);
 114                 }
 115
 116                 $ch = @curl_init($url);
 117
 118                 if (($redirects > 8) || (!$ch)) {
 119                         return CurlResult::createErrorCurl($url);
 120                 }
 121
 122                 @curl_setopt($ch, CURLOPT_HEADER, true);
 123
 124                 if (!empty($opts['cookiejar'])) {
 125                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 126                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 127                 }
 128
 129                 // These settings aren't needed. We're following the location already.
 130                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 131                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 132
 133                 if (!empty($opts['accept_content'])) {
 134                         curl_setopt(
 135                                 $ch,
 136                                 CURLOPT_HTTPHEADER,
 137                                 ['Accept: ' . $opts['accept_content']]
 138                         );
 139                 }
 140
 141                 if (!empty($opts['header'])) {
 142                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 143                 }
 144
 145                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 146                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 147
 148                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 149
 150                 if ($range > 0) {
 151                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 152                 }
 153
 154                 // Without this setting it seems as if some webservers send compressed content
 155                 // This seems to confuse curl so that it shows this uncompressed.
 156                 /// @todo  We could possibly set this value to "gzip" or something similar
 157                 curl_setopt($ch, CURLOPT_ENCODING, '');
 158
 159                 if (!empty($opts['headers'])) {
 160                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 161                 }
 162
 163                 if (!empty($opts['nobody'])) {
 164                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 165                 }
 166
 167                 if (!empty($opts['timeout'])) {
 168                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 169                 } else {
 170                         $curl_time = Config::get('system', 'curl_timeout', 60);
 171                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 172                 }
 173
 174                 // by default we will allow self-signed certs
 175                 // but you can override this
 176
 177                 $check_cert = Config::get('system', 'verifyssl');
 178                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 179
 180                 if ($check_cert) {
 181                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 182                 }
 183
 184                 $proxy = Config::get('system', 'proxy');
 185
 186                 if (strlen($proxy)) {
 187                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 188                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 189                         $proxyuser = @Config::get('system', 'proxyuser');
 190
 191                         if (strlen($proxyuser)) {
 192                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 193                         }
 194                 }
 195
 196                 if (Config::get('system', 'ipv4_resolve', false)) {
 197                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 198                 }
 199
 200                 if ($binary) {
 201                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 202                 }
 203
 204                 // don't let curl abort the entire application
 205                 // if it throws any errors.
 206
 207                 $s = @curl_exec($ch);
 208                 $curl_info = @curl_getinfo($ch);
 209
 210                 // Special treatment for HTTP Code 416
 211                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 212                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 213                         @curl_setopt($ch, CURLOPT_RANGE, '');
 214                         $s = @curl_exec($ch);
 215                         $curl_info = @curl_getinfo($ch);
 216                 }
 217
 218                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 219
 220                 if ($curlResponse->isRedirectUrl()) {
 221                         $redirects++;
 222                         Logger::log('curl: redirect ' . $url . ' to ' . $curlResponse->getRedirectUrl());
 223                         @curl_close($ch);
 224                         return self::curl($curlResponse->getRedirectUrl(), $binary, $redirects, $opts);
 225                 }
 226
 227                 @curl_close($ch);
 228
 229                 $a->saveTimestamp($stamp1, 'network');
 230
 231                 return $curlResponse;
 232         }
 233
 234         /**
 235          * @brief Send POST request to $url
 236          *
 237          * @param string  $url       URL to post
 238          * @param mixed   $params    array of POST variables
 239          * @param string  $headers   HTTP headers
 240          * @param integer $redirects Recursion counter for internal use - default = 0
 241          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 242          *
 243          * @return CurlResult The content
 244          */
 245         public static function post($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 246         {
 247                 $stamp1 = microtime(true);
 248
 249                 if (self::isUrlBlocked($url)) {
 250                         Logger::log('post_url: domain of ' . $url . ' is blocked', Logger::DATA);
 251                         return CurlResult::createErrorCurl($url);
 252                 }
 253
 254                 $a = get_app();
 255                 $ch = curl_init($url);
 256
 257                 if (($redirects > 8) || (!$ch)) {
 258                         return CurlResult::createErrorCurl($url);
 259                 }
 260
 261                 Logger::log('post_url: start ' . $url, Logger::DATA);
 262
 263                 curl_setopt($ch, CURLOPT_HEADER, true);
 264                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 265                 curl_setopt($ch, CURLOPT_POST, 1);
 266                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 267                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 268
 269                 if (Config::get('system', 'ipv4_resolve', false)) {
 270                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 271                 }
 272
 273                 if (intval($timeout)) {
 274                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 275                 } else {
 276                         $curl_time = Config::get('system', 'curl_timeout', 60);
 277                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 278                 }
 279
 280                 if (defined('LIGHTTPD')) {
 281                         if (!is_array($headers)) {
 282                                 $headers = ['Expect:'];
 283                         } else {
 284                                 if (!in_array('Expect:', $headers)) {
 285                                         array_push($headers, 'Expect:');
 286                                 }
 287                         }
 288                 }
 289
 290                 if ($headers) {
 291                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 292                 }
 293
 294                 $check_cert = Config::get('system', 'verifyssl');
 295                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 296
 297                 if ($check_cert) {
 298                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 299                 }
 300
 301                 $proxy = Config::get('system', 'proxy');
 302
 303                 if (strlen($proxy)) {
 304                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 305                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 306                         $proxyuser = Config::get('system', 'proxyuser');
 307                         if (strlen($proxyuser)) {
 308                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 309                         }
 310                 }
 311
 312                 // don't let curl abort the entire application
 313                 // if it throws any errors.
 314
 315                 $s = @curl_exec($ch);
 316
 317                 $base = $s;
 318                 $curl_info = curl_getinfo($ch);
 319
 320                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 321
 322                 if ($curlResponse->isRedirectUrl()) {
 323                         $redirects++;
 324                         Logger::log('post_url: redirect ' . $url . ' to ' . $curlResponse->getRedirectUrl());
 325                         curl_close($ch);
 326                         return self::post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 327                 }
 328
 329                 curl_close($ch);
 330
 331                 $a->saveTimestamp($stamp1, 'network');
 332
 333                 Logger::log('post_url: end ' . $url, Logger::DATA);
 334
 335                 return $curlResponse;
 336         }
 337
 338         /**
 339          * @brief Check URL to see if it's real
 340          *
 341          * Take a URL from the wild, prepend http:// if necessary
 342          * and check DNS to see if it's real (or check if is a valid IP address)
 343          *
 344          * @param string $url The URL to be validated
 345          * @return string|boolean The actual working URL, false else
 346          */
 347         public static function isUrlValid($url)
 348         {
 349                 if (Config::get('system', 'disable_url_validation')) {
 350                         return $url;
 351                 }
 352
 353                 // no naked subdomains (allow localhost for tests)
 354                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 355                         return false;
 356                 }
 357
 358                 if (substr($url, 0, 4) != 'http') {
 359                         $url = 'http://' . $url;
 360                 }
 361
 362                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 363                 $h = @parse_url($url);
 364
 365                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 366                         return $url;
 367                 }
 368
 369                 return false;
 370         }
 371
 372         /**
 373          * @brief Checks that email is an actual resolvable internet address
 374          *
 375          * @param string $addr The email address
 376          * @return boolean True if it's a valid email address, false if it's not
 377          */
 378         public static function isEmailDomainValid($addr)
 379         {
 380                 if (Config::get('system', 'disable_email_validation')) {
 381                         return true;
 382                 }
 383
 384                 if (! strpos($addr, '@')) {
 385                         return false;
 386                 }
 387
 388                 $h = substr($addr, strpos($addr, '@') + 1);
 389
 390                 // Concerning the @ see here: https://stackoverflow.com/questions/36280957/dns-get-record-a-temporary-server-error-occurred
 391                 if ($h && (@dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 392                         return true;
 393                 }
 394                 if ($h && @dns_get_record($h, DNS_CNAME + DNS_MX)) {
 395                         return true;
 396                 }
 397                 return false;
 398         }
 399
 400         /**
 401          * @brief Check if URL is allowed
 402          *
 403          * Check $url against our list of allowed sites,
 404          * wildcards allowed. If allowed_sites is unset return true;
 405          *
 406          * @param string $url URL which get tested
 407          * @return boolean True if url is allowed otherwise return false
 408          */
 409         public static function isUrlAllowed($url)
 410         {
 411                 $h = @parse_url($url);
 412
 413                 if (! $h) {
 414                         return false;
 415                 }
 416
 417                 $str_allowed = Config::get('system', 'allowed_sites');
 418                 if (! $str_allowed) {
 419                         return true;
 420                 }
 421
 422                 $found = false;
 423
 424                 $host = strtolower($h['host']);
 425
 426                 // always allow our own site
 427                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 428                         return true;
 429                 }
 430
 431                 $fnmatch = function_exists('fnmatch');
 432                 $allowed = explode(',', $str_allowed);
 433
 434                 if (count($allowed)) {
 435                         foreach ($allowed as $a) {
 436                                 $pat = strtolower(trim($a));
 437                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 438                                         $found = true;
 439                                         break;
 440                                 }
 441                         }
 442                 }
 443                 return $found;
 444         }
 445
 446         /**
 447          * Checks if the provided url domain is on the domain blocklist.
 448          * Returns true if it is or malformed URL, false if not.
 449          *
 450          * @param string $url The url to check the domain from
 451          *
 452          * @return boolean
 453          */
 454         public static function isUrlBlocked($url)
 455         {
 456                 $host = @parse_url($url, PHP_URL_HOST);
 457                 if (!$host) {
 458                         return false;
 459                 }
 460
 461                 $domain_blocklist = Config::get('system', 'blocklist', []);
 462                 if (!$domain_blocklist) {
 463                         return false;
 464                 }
 465
 466                 foreach ($domain_blocklist as $domain_block) {
 467                         if (strcasecmp($domain_block['domain'], $host) === 0) {
 468                                 return true;
 469                         }
 470                 }
 471
 472                 return false;
 473         }
 474
 475         /**
 476          * @brief Check if email address is allowed to register here.
 477          *
 478          * Compare against our list (wildcards allowed).
 479          *
 480          * @param  string $email email address
 481          * @return boolean False if not allowed, true if allowed
 482          *    or if allowed list is not configured
 483          */
 484         public static function isEmailDomainAllowed($email)
 485         {
 486                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 487                 if (!$domain) {
 488                         return false;
 489                 }
 490
 491                 $str_allowed = Config::get('system', 'allowed_email', '');
 492                 if (empty($str_allowed)) {
 493                         return true;
 494                 }
 495
 496                 $allowed = explode(',', $str_allowed);
 497
 498                 return self::isDomainAllowed($domain, $allowed);
 499         }
 500
 501         /**
 502          * Checks for the existence of a domain in a domain list
 503          *
 504          * @brief Checks for the existence of a domain in a domain list
 505          * @param string $domain
 506          * @param array  $domain_list
 507          * @return boolean
 508          */
 509         public static function isDomainAllowed($domain, array $domain_list)
 510         {
 511                 $found = false;
 512
 513                 foreach ($domain_list as $item) {
 514                         $pat = strtolower(trim($item));
 515                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 516                                 $found = true;
 517                                 break;
 518                         }
 519                 }
 520
 521                 return $found;
 522         }
 523
 524         public static function lookupAvatarByEmail($email)
 525         {
 526                 $avatar['size'] = 300;
 527                 $avatar['email'] = $email;
 528                 $avatar['url'] = '';
 529                 $avatar['success'] = false;
 530
 531                 Addon::callHooks('avatar_lookup', $avatar);
 532
 533                 if (! $avatar['success']) {
 534                         $avatar['url'] = System::baseUrl() . '/images/person-300.jpg';
 535                 }
 536
 537                 Logger::log('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], Logger::DEBUG);
 538                 return $avatar['url'];
 539         }
 540
 541         /**
 542          * @brief Remove Google Analytics and other tracking platforms params from URL
 543          *
 544          * @param string $url Any user-submitted URL that may contain tracking params
 545          * @return string The same URL stripped of tracking parameters
 546          */
 547         public static function stripTrackingQueryParams($url)
 548         {
 549                 $urldata = parse_url($url);
 550                 if (!empty($urldata["query"])) {
 551                         $query = $urldata["query"];
 552                         parse_str($query, $querydata);
 553
 554                         if (is_array($querydata)) {
 555                                 foreach ($querydata as $param => $value) {
 556                                         if (in_array(
 557                                                 $param,
 558                                                 [
 559                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 560                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 561                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 562                                                         "awesm", "wtrid",
 563                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 564                                                 )
 565                                         ) {
 566                                                 $pair = $param . "=" . urlencode($value);
 567                                                 $url = str_replace($pair, "", $url);
 568
 569                                                 // Second try: if the url isn't encoded completely
 570                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 571                                                 $url = str_replace($pair, "", $url);
 572
 573                                                 // Third try: Maybey the url isn't encoded at all
 574                                                 $pair = $param . "=" . $value;
 575                                                 $url = str_replace($pair, "", $url);
 576
 577                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 578                                         }
 579                                 }
 580                         }
 581
 582                         if (substr($url, -1, 1) == "?") {
 583                                 $url = substr($url, 0, -1);
 584                         }
 585                 }
 586
 587                 return $url;
 588         }
 589
 590         /**
 591          * @brief Returns the original URL of the provided URL
 592          *
 593          * This function strips tracking query params and follows redirections, either
 594          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 595          *
 596          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 597          *
 598          * @see ParseUrl::getSiteinfo
 599          *
 600          * @param string $url       A user-submitted URL
 601          * @param int    $depth     The current redirection recursion level (internal)
 602          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 603          * @return string A canonical URL
 604          */
 605         public static function finalUrl($url, $depth = 1, $fetchbody = false)
 606         {
 607                 $a = get_app();
 608
 609                 $url = self::stripTrackingQueryParams($url);
 610
 611                 if ($depth > 10) {
 612                         return $url;
 613                 }
 614
 615                 $url = trim($url, "'");
 616
 617                 $stamp1 = microtime(true);
 618
 619                 $ch = curl_init();
 620                 curl_setopt($ch, CURLOPT_URL, $url);
 621                 curl_setopt($ch, CURLOPT_HEADER, 1);
 622                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 623                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 624                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 625                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 626
 627                 curl_exec($ch);
 628                 $curl_info = @curl_getinfo($ch);
 629                 $http_code = $curl_info['http_code'];
 630                 curl_close($ch);
 631
 632                 $a->saveTimestamp($stamp1, "network");
 633
 634                 if ($http_code == 0) {
 635                         return $url;
 636                 }
 637
 638                 if (in_array($http_code, ['301', '302'])) {
 639                         if (!empty($curl_info['redirect_url'])) {
 640                                 return self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 641                         } elseif (!empty($curl_info['location'])) {
 642                                 return self::finalUrl($curl_info['location'], ++$depth, $fetchbody);
 643                         }
 644                 }
 645
 646                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 647                 if (!$fetchbody) {
 648                         return(self::finalUrl($url, ++$depth, true));
 649                 }
 650
 651                 // if the file is too large then exit
 652                 if ($curl_info["download_content_length"] > 1000000) {
 653                         return $url;
 654                 }
 655
 656                 // if it isn't a HTML file then exit
 657                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 658                         return $url;
 659                 }
 660
 661                 $stamp1 = microtime(true);
 662
 663                 $ch = curl_init();
 664                 curl_setopt($ch, CURLOPT_URL, $url);
 665                 curl_setopt($ch, CURLOPT_HEADER, 0);
 666                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 667                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 668                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 669                 curl_setopt($ch, CURLOPT_USERAGENT, $a->getUserAgent());
 670
 671                 $body = curl_exec($ch);
 672                 curl_close($ch);
 673
 674                 $a->saveTimestamp($stamp1, "network");
 675
 676                 if (trim($body) == "") {
 677                         return $url;
 678                 }
 679
 680                 // Check for redirect in meta elements
 681                 $doc = new DOMDocument();
 682                 @$doc->loadHTML($body);
 683
 684                 $xpath = new DomXPath($doc);
 685
 686                 $list = $xpath->query("//meta[@content]");
 687                 foreach ($list as $node) {
 688                         $attr = [];
 689                         if ($node->attributes->length) {
 690                                 foreach ($node->attributes as $attribute) {
 691                                         $attr[$attribute->name] = $attribute->value;
 692                                 }
 693                         }
 694
 695                         if (@$attr["http-equiv"] == 'refresh') {
 696                                 $path = $attr["content"];
 697                                 $pathinfo = explode(";", $path);
 698                                 foreach ($pathinfo as $value) {
 699                                         if (substr(strtolower($value), 0, 4) == "url=") {
 700                                                 return self::finalUrl(substr($value, 4), ++$depth);
 701                                         }
 702                                 }
 703                         }
 704                 }
 705
 706                 return $url;
 707         }
 708
 709         /**
 710          * @brief Find the matching part between two url
 711          *
 712          * @param string $url1
 713          * @param string $url2
 714          * @return string The matching part
 715          */
 716         public static function getUrlMatch($url1, $url2)
 717         {
 718                 if (($url1 == "") || ($url2 == "")) {
 719                         return "";
 720                 }
 721
 722                 $url1 = Strings::normaliseLink($url1);
 723                 $url2 = Strings::normaliseLink($url2);
 724
 725                 $parts1 = parse_url($url1);
 726                 $parts2 = parse_url($url2);
 727
 728                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 729                         return "";
 730                 }
 731
 732                 if (empty($parts1["scheme"])) {
 733                         $parts1["scheme"] = '';
 734                 }
 735                 if (empty($parts2["scheme"])) {
 736                         $parts2["scheme"] = '';
 737                 }
 738
 739                 if ($parts1["scheme"] != $parts2["scheme"]) {
 740                         return "";
 741                 }
 742
 743                 if (empty($parts1["host"])) {
 744                         $parts1["host"] = '';
 745                 }
 746                 if (empty($parts2["host"])) {
 747                         $parts2["host"] = '';
 748                 }
 749
 750                 if ($parts1["host"] != $parts2["host"]) {
 751                         return "";
 752                 }
 753
 754                 if (empty($parts1["port"])) {
 755                         $parts1["port"] = '';
 756                 }
 757                 if (empty($parts2["port"])) {
 758                         $parts2["port"] = '';
 759                 }
 760
 761                 if ($parts1["port"] != $parts2["port"]) {
 762                         return "";
 763                 }
 764
 765                 $match = $parts1["scheme"]."://".$parts1["host"];
 766
 767                 if ($parts1["port"]) {
 768                         $match .= ":".$parts1["port"];
 769                 }
 770
 771                 if (empty($parts1["path"])) {
 772                         $parts1["path"] = '';
 773                 }
 774                 if (empty($parts2["path"])) {
 775                         $parts2["path"] = '';
 776                 }
 777
 778                 $pathparts1 = explode("/", $parts1["path"]);
 779                 $pathparts2 = explode("/", $parts2["path"]);
 780
 781                 $i = 0;
 782                 $path = "";
 783                 do {
 784                         $path1 = defaults($pathparts1, $i, '');
 785                         $path2 = defaults($pathparts2, $i, '');
 786
 787                         if ($path1 == $path2) {
 788                                 $path .= $path1."/";
 789                         }
 790                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 791
 792                 $match .= $path;
 793
 794                 return Strings::normaliseLink($match);
 795         }
 796
 797         /**
 798          * @brief Glue url parts together
 799          *
 800          * @param array $parsed URL parts
 801          *
 802          * @return string The glued URL
 803          */
 804         public static function unparseURL($parsed)
 805         {
 806                 $get = function ($key) use ($parsed) {
 807                         return isset($parsed[$key]) ? $parsed[$key] : null;
 808                 };
 809
 810                 $pass      = $get('pass');
 811                 $user      = $get('user');
 812                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 813                 $port      = $get('port');
 814                 $scheme    = $get('scheme');
 815                 $query     = $get('query');
 816                 $fragment  = $get('fragment');
 817                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 818                                                 $get('host') .
 819                                                 ($port ? ":$port" : '');
 820
 821                 return  (strlen($scheme) ? $scheme.":" : '') .
 822                         (strlen($authority) ? "//".$authority : '') .
 823                         $get('path') .
 824                         (strlen($query) ? "?".$query : '') .
 825                         (strlen($fragment) ? "#".$fragment : '');
 826         }
 827 }