src/Util/Network.php

   1 <?php
   2 /**
   3  * @file src/Util/Network.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use Friendica\App;
   8 use Friendica\Core\Addon;
   9 use Friendica\Core\L10n;
  10 use Friendica\Core\System;
  11 use Friendica\Core\Config;
  12 use Friendica\Network\Probe;
  13 use Friendica\Object\Image;
  14 use Friendica\Util\XML;
  15 use DOMDocument;
  16 use DomXPath;
  17
  18 class Network
  19 {
  20         /**
  21          * @brief Curl wrapper
  22          *
  23          * If binary flag is true, return binary results.
  24          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
  25          * to preserve cookies from one request to the next.
  26          *
  27          * @param string  $url            URL to fetch
  28          * @param boolean $binary         default false
  29          *                                TRUE if asked to return binary results (file download)
  30          * @param integer $redirects      The recursion counter for internal use - default 0
  31          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  32          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  33          * @param string  $cookiejar      Path to cookie jar file
  34          *
  35          * @return string The fetched content
  36          */
  37         public static function fetchUrl($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  38         {
  39                 $ret = self::curl(
  40                         $url,
  41                         $binary,
  42                         $redirects,
  43                         ['timeout'=>$timeout,
  44                         'accept_content'=>$accept_content,
  45                         'cookiejar'=>$cookiejar
  46                         ]
  47                 );
  48
  49                 return($ret['body']);
  50         }
  51
  52         /**
  53          * @brief fetches an URL.
  54          *
  55          * @param string  $url       URL to fetch
  56          * @param boolean $binary    default false
  57          *                           TRUE if asked to return binary results (file download)
  58          * @param int     $redirects The recursion counter for internal use - default 0
  59          * @param array   $opts      (optional parameters) assoziative array with:
  60          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  61          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  62          *                           'http_auth' => username:password
  63          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  64          *                           'nobody' => only return the header
  65          *                           'cookiejar' => path to cookie jar file
  66          *
  67          * @return array an assoziative array with:
  68          *    int 'return_code' => HTTP return code or 0 if timeout or failure
  69          *    boolean 'success' => boolean true (if HTTP 2xx result) or false
  70          *    string 'redirect_url' => in case of redirect, content was finally retrieved from this URL
  71          *    string 'header' => HTTP headers
  72          *    string 'body' => fetched content
  73          */
  74         public static function curl($url, $binary = false, &$redirects = 0, $opts = [])
  75         {
  76                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
  77
  78                 $stamp1 = microtime(true);
  79
  80                 $a = get_app();
  81
  82                 if (self::isUrlBlocked($url)) {
  83                         logger('domain of ' . $url . ' is blocked', LOGGER_DATA);
  84                         return $ret;
  85                 }
  86
  87                 $ch = @curl_init($url);
  88
  89                 if (($redirects > 8) || (!$ch)) {
  90                         return $ret;
  91                 }
  92
  93                 @curl_setopt($ch, CURLOPT_HEADER, true);
  94
  95                 if (x($opts, "cookiejar")) {
  96                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
  97                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
  98                 }
  99
 100                 // These settings aren't needed. We're following the location already.
 101                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 102                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 103
 104                 if (x($opts, 'accept_content')) {
 105                         curl_setopt(
 106                                 $ch,
 107                                 CURLOPT_HTTPHEADER,
 108                                 ['Accept: ' . $opts['accept_content']]
 109                         );
 110                 }
 111
 112                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 113                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 114
 115                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 116
 117                 if ($range > 0) {
 118                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 119                 }
 120
 121                 // Without this setting it seems as if some webservers send compressed content
 122                 // This seems to confuse curl so that it shows this uncompressed.
 123                 /// @todo  We could possibly set this value to "gzip" or something similar
 124                 curl_setopt($ch, CURLOPT_ENCODING, '');
 125
 126                 if (x($opts, 'headers')) {
 127                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 128                 }
 129
 130                 if (x($opts, 'nobody')) {
 131                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 132                 }
 133
 134                 if (x($opts, 'timeout')) {
 135                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 136                 } else {
 137                         $curl_time = Config::get('system', 'curl_timeout', 60);
 138                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 139                 }
 140
 141                 // by default we will allow self-signed certs
 142                 // but you can override this
 143
 144                 $check_cert = Config::get('system', 'verifyssl');
 145                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 146
 147                 if ($check_cert) {
 148                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 149                 }
 150
 151                 $proxy = Config::get('system', 'proxy');
 152
 153                 if (strlen($proxy)) {
 154                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 155                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 156                         $proxyuser = @Config::get('system', 'proxyuser');
 157
 158                         if (strlen($proxyuser)) {
 159                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 160                         }
 161                 }
 162
 163                 if (Config::get('system', 'ipv4_resolve', false)) {
 164                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 165                 }
 166
 167                 if ($binary) {
 168                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 169                 }
 170
 171                 $a->set_curl_code(0);
 172
 173                 // don't let curl abort the entire application
 174                 // if it throws any errors.
 175
 176                 $s = @curl_exec($ch);
 177                 $curl_info = @curl_getinfo($ch);
 178
 179                 // Special treatment for HTTP Code 416
 180                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 181                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 182                         @curl_setopt($ch, CURLOPT_RANGE, '');
 183                         $s = @curl_exec($ch);
 184                         $curl_info = @curl_getinfo($ch);
 185                 }
 186
 187                 if (curl_errno($ch) !== CURLE_OK) {
 188                         logger('error fetching ' . $url . ': ' . curl_error($ch), LOGGER_NORMAL);
 189                 }
 190
 191                 $ret['errno'] = curl_errno($ch);
 192
 193                 $base = $s;
 194                 $ret['info'] = $curl_info;
 195
 196                 $http_code = $curl_info['http_code'];
 197
 198                 logger($url . ': ' . $http_code . " " . $s, LOGGER_DATA);
 199                 $header = '';
 200
 201                 // Pull out multiple headers, e.g. proxy and continuation headers
 202                 // allow for HTTP/2.x without fixing code
 203
 204                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 205                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 206                         $header .= $chunk;
 207                         $base = substr($base, strlen($chunk));
 208                 }
 209
 210                 $a->set_curl_code($http_code);
 211                 $a->set_curl_content_type($curl_info['content_type']);
 212                 $a->set_curl_headers($header);
 213
 214                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 215                         $new_location_info = @parse_url($curl_info['redirect_url']);
 216                         $old_location_info = @parse_url($curl_info['url']);
 217
 218                         $newurl = $curl_info['redirect_url'];
 219
 220                         if (($new_location_info['path'] == '') && ( $new_location_info['host'] != '')) {
 221                                 $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
 222                         }
 223
 224                         $matches = [];
 225
 226                         if (preg_match('/(Location:|URI:)(.*?)\n/i', $header, $matches)) {
 227                                 $newurl = trim(array_pop($matches));
 228                         }
 229                         if (strpos($newurl, '/') === 0) {
 230                                 $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
 231                         }
 232
 233                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 234                                 $redirects++;
 235                                 @curl_close($ch);
 236                                 return self::curl($newurl, $binary, $redirects, $opts);
 237                         }
 238                 }
 239
 240                 $a->set_curl_code($http_code);
 241                 $a->set_curl_content_type($curl_info['content_type']);
 242
 243                 $rc = intval($http_code);
 244                 $ret['return_code'] = $rc;
 245                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 246                 $ret['redirect_url'] = $url;
 247
 248                 if (!$ret['success']) {
 249                         $ret['error'] = curl_error($ch);
 250                         $ret['debug'] = $curl_info;
 251                         logger('error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 252                         logger('debug: '.print_r($curl_info, true), LOGGER_DATA);
 253                 }
 254
 255                 $ret['body'] = substr($s, strlen($header));
 256                 $ret['header'] = $header;
 257
 258                 if (x($opts, 'debug')) {
 259                         $ret['debug'] = $curl_info;
 260                 }
 261
 262                 @curl_close($ch);
 263
 264                 $a->save_timestamp($stamp1, 'network');
 265
 266                 return($ret);
 267         }
 268
 269         /**
 270          * @brief Send POST request to $url
 271          *
 272          * @param string  $url       URL to post
 273          * @param mixed   $params    array of POST variables
 274          * @param string  $headers   HTTP headers
 275          * @param integer $redirects Recursion counter for internal use - default = 0
 276          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 277          *
 278          * @return string The content
 279          */
 280         public static function post($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 281         {
 282                 $stamp1 = microtime(true);
 283
 284                 if (self::isUrlBlocked($url)) {
 285                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 286                         return false;
 287                 }
 288
 289                 $a = get_app();
 290                 $ch = curl_init($url);
 291
 292                 if (($redirects > 8) || (!$ch)) {
 293                         return false;
 294                 }
 295
 296                 logger('post_url: start ' . $url, LOGGER_DATA);
 297
 298                 curl_setopt($ch, CURLOPT_HEADER, true);
 299                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 300                 curl_setopt($ch, CURLOPT_POST, 1);
 301                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 302                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 303
 304                 if (Config::get('system', 'ipv4_resolve', false)) {
 305                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 306                 }
 307
 308                 if (intval($timeout)) {
 309                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 310                 } else {
 311                         $curl_time = Config::get('system', 'curl_timeout', 60);
 312                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 313                 }
 314
 315                 if (defined('LIGHTTPD')) {
 316                         if (!is_array($headers)) {
 317                                 $headers = ['Expect:'];
 318                         } else {
 319                                 if (!in_array('Expect:', $headers)) {
 320                                         array_push($headers, 'Expect:');
 321                                 }
 322                         }
 323                 }
 324
 325                 if ($headers) {
 326                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 327                 }
 328
 329                 $check_cert = Config::get('system', 'verifyssl');
 330                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 331
 332                 if ($check_cert) {
 333                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 334                 }
 335
 336                 $proxy = Config::get('system', 'proxy');
 337
 338                 if (strlen($proxy)) {
 339                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 340                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 341                         $proxyuser = Config::get('system', 'proxyuser');
 342                         if (strlen($proxyuser)) {
 343                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 344                         }
 345                 }
 346
 347                 $a->set_curl_code(0);
 348
 349                 // don't let curl abort the entire application
 350                 // if it throws any errors.
 351
 352                 $s = @curl_exec($ch);
 353
 354                 $base = $s;
 355                 $curl_info = curl_getinfo($ch);
 356                 $http_code = $curl_info['http_code'];
 357
 358                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 359
 360                 $header = '';
 361
 362                 // Pull out multiple headers, e.g. proxy and continuation headers
 363                 // allow for HTTP/2.x without fixing code
 364
 365                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 366                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 367                         $header .= $chunk;
 368                         $base = substr($base, strlen($chunk));
 369                 }
 370
 371                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 372                         $matches = [];
 373                         $new_location_info = @parse_url($curl_info['redirect_url']);
 374                         $old_location_info = @parse_url($curl_info['url']);
 375
 376                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 377                         $newurl = trim(array_pop($matches));
 378
 379                         if (strpos($newurl, '/') === 0) {
 380                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 381                         }
 382
 383                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 384                                 $redirects++;
 385                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 386                                 return self::post($newurl, $params, $headers, $redirects, $timeout);
 387                         }
 388                 }
 389
 390                 $a->set_curl_code($http_code);
 391
 392                 $body = substr($s, strlen($header));
 393
 394                 $a->set_curl_headers($header);
 395
 396                 curl_close($ch);
 397
 398                 $a->save_timestamp($stamp1, 'network');
 399
 400                 logger('post_url: end ' . $url, LOGGER_DATA);
 401
 402                 return $body;
 403         }
 404
 405         /**
 406          * @brief Check URL to see if it's real
 407          *
 408          * Take a URL from the wild, prepend http:// if necessary
 409          * and check DNS to see if it's real (or check if is a valid IP address)
 410          *
 411          * @param string $url The URL to be validated
 412          * @return string|boolean The actual working URL, false else
 413          */
 414         public static function isUrlValid($url)
 415         {
 416                 if (Config::get('system', 'disable_url_validation')) {
 417                         return $url;
 418                 }
 419
 420                 // no naked subdomains (allow localhost for tests)
 421                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 422                         return false;
 423                 }
 424
 425                 if (substr($url, 0, 4) != 'http') {
 426                         $url = 'http://' . $url;
 427                 }
 428
 429                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 430                 $h = @parse_url($url);
 431
 432                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 433                         return $url;
 434                 }
 435
 436                 return false;
 437         }
 438
 439         /**
 440          * @brief Checks that email is an actual resolvable internet address
 441          *
 442          * @param string $addr The email address
 443          * @return boolean True if it's a valid email address, false if it's not
 444          */
 445         public static function isEmailDomainValid($addr)
 446         {
 447                 if (Config::get('system', 'disable_email_validation')) {
 448                         return true;
 449                 }
 450
 451                 if (! strpos($addr, '@')) {
 452                         return false;
 453                 }
 454
 455                 $h = substr($addr, strpos($addr, '@') + 1);
 456
 457                 if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 458                         return true;
 459                 }
 460                 return false;
 461         }
 462
 463         /**
 464          * @brief Check if URL is allowed
 465          *
 466          * Check $url against our list of allowed sites,
 467          * wildcards allowed. If allowed_sites is unset return true;
 468          *
 469          * @param string $url URL which get tested
 470          * @return boolean True if url is allowed otherwise return false
 471          */
 472         public static function isUrlAllowed($url)
 473         {
 474                 $h = @parse_url($url);
 475
 476                 if (! $h) {
 477                         return false;
 478                 }
 479
 480                 $str_allowed = Config::get('system', 'allowed_sites');
 481                 if (! $str_allowed) {
 482                         return true;
 483                 }
 484
 485                 $found = false;
 486
 487                 $host = strtolower($h['host']);
 488
 489                 // always allow our own site
 490                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 491                         return true;
 492                 }
 493
 494                 $fnmatch = function_exists('fnmatch');
 495                 $allowed = explode(',', $str_allowed);
 496
 497                 if (count($allowed)) {
 498                         foreach ($allowed as $a) {
 499                                 $pat = strtolower(trim($a));
 500                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 501                                         $found = true;
 502                                         break;
 503                                 }
 504                         }
 505                 }
 506                 return $found;
 507         }
 508
 509         /**
 510          * Checks if the provided url domain is on the domain blocklist.
 511          * Returns true if it is or malformed URL, false if not.
 512          *
 513          * @param string $url The url to check the domain from
 514          *
 515          * @return boolean
 516          */
 517         public static function isUrlBlocked($url)
 518         {
 519                 $h = @parse_url($url);
 520
 521                 if (! $h) {
 522                         return true;
 523                 }
 524
 525                 $domain_blocklist = Config::get('system', 'blocklist', []);
 526                 if (! $domain_blocklist) {
 527                         return false;
 528                 }
 529
 530                 $host = strtolower($h['host']);
 531
 532                 foreach ($domain_blocklist as $domain_block) {
 533                         if (strtolower($domain_block['domain']) == $host) {
 534                                 return true;
 535                         }
 536                 }
 537
 538                 return false;
 539         }
 540
 541         /**
 542          * @brief Check if email address is allowed to register here.
 543          *
 544          * Compare against our list (wildcards allowed).
 545          *
 546          * @param  string $email email address
 547          * @return boolean False if not allowed, true if allowed
 548          *    or if allowed list is not configured
 549          */
 550         public static function isEmailDomainAllowed($email)
 551         {
 552                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 553                 if (!$domain) {
 554                         return false;
 555                 }
 556
 557                 $str_allowed = Config::get('system', 'allowed_email', '');
 558                 if (!x($str_allowed)) {
 559                         return true;
 560                 }
 561
 562                 $allowed = explode(',', $str_allowed);
 563
 564                 return self::isDomainAllowed($domain, $allowed);
 565         }
 566
 567         /**
 568          * Checks for the existence of a domain in a domain list
 569          *
 570          * @brief Checks for the existence of a domain in a domain list
 571          * @param string $domain
 572          * @param array  $domain_list
 573          * @return boolean
 574          */
 575         public static function isDomainAllowed($domain, array $domain_list)
 576         {
 577                 $found = false;
 578
 579                 foreach ($domain_list as $item) {
 580                         $pat = strtolower(trim($item));
 581                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 582                                 $found = true;
 583                                 break;
 584                         }
 585                 }
 586
 587                 return $found;
 588         }
 589
 590         public static function lookupAvatarByEmail($email)
 591         {
 592                 $avatar['size'] = 175;
 593                 $avatar['email'] = $email;
 594                 $avatar['url'] = '';
 595                 $avatar['success'] = false;
 596
 597                 Addon::callHooks('avatar_lookup', $avatar);
 598
 599                 if (! $avatar['success']) {
 600                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 601                 }
 602
 603                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 604                 return $avatar['url'];
 605         }
 606
 607         /**
 608          * @brief Remove Google Analytics and other tracking platforms params from URL
 609          *
 610          * @param string $url Any user-submitted URL that may contain tracking params
 611          * @return string The same URL stripped of tracking parameters
 612          */
 613         public static function stripTrackingQueryParams($url)
 614         {
 615                 $urldata = parse_url($url);
 616                 if (is_string($urldata["query"])) {
 617                         $query = $urldata["query"];
 618                         parse_str($query, $querydata);
 619
 620                         if (is_array($querydata)) {
 621                                 foreach ($querydata as $param => $value) {
 622                                         if (in_array(
 623                                                 $param,
 624                                                 [
 625                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 626                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 627                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 628                                                         "awesm", "wtrid",
 629                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 630                                                 )
 631                                         ) {
 632                                                 $pair = $param . "=" . urlencode($value);
 633                                                 $url = str_replace($pair, "", $url);
 634
 635                                                 // Second try: if the url isn't encoded completely
 636                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 637                                                 $url = str_replace($pair, "", $url);
 638
 639                                                 // Third try: Maybey the url isn't encoded at all
 640                                                 $pair = $param . "=" . $value;
 641                                                 $url = str_replace($pair, "", $url);
 642
 643                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 644                                         }
 645                                 }
 646                         }
 647
 648                         if (substr($url, -1, 1) == "?") {
 649                                 $url = substr($url, 0, -1);
 650                         }
 651                 }
 652
 653                 return $url;
 654         }
 655
 656         /**
 657          * @brief Returns the original URL of the provided URL
 658          *
 659          * This function strips tracking query params and follows redirections, either
 660          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 661          *
 662          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 663          *
 664          * @see ParseUrl::getSiteinfo
 665          *
 666          * @param string $url       A user-submitted URL
 667          * @param int    $depth     The current redirection recursion level (internal)
 668          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 669          * @return string A canonical URL
 670          */
 671         public static function finalUrl($url, $depth = 1, $fetchbody = false)
 672         {
 673                 $a = get_app();
 674
 675                 $url = self::stripTrackingQueryParams($url);
 676
 677                 if ($depth > 10) {
 678                         return($url);
 679                 }
 680
 681                 $url = trim($url, "'");
 682
 683                 $stamp1 = microtime(true);
 684
 685                 $ch = curl_init();
 686                 curl_setopt($ch, CURLOPT_URL, $url);
 687                 curl_setopt($ch, CURLOPT_HEADER, 1);
 688                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 689                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 690                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 691                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 692
 693                 curl_exec($ch);
 694                 $curl_info = @curl_getinfo($ch);
 695                 $http_code = $curl_info['http_code'];
 696                 curl_close($ch);
 697
 698                 $a->save_timestamp($stamp1, "network");
 699
 700                 if ($http_code == 0) {
 701                         return($url);
 702                 }
 703
 704                 if ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 705                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 706                 ) {
 707                         if ($curl_info['redirect_url'] != "") {
 708                                 return(self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody));
 709                         } else {
 710                                 return(self::finalUrl($curl_info['location'], ++$depth, $fetchbody));
 711                         }
 712                 }
 713
 714                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 715                 if (!$fetchbody) {
 716                         return(self::finalUrl($url, ++$depth, true));
 717                 }
 718
 719                 // if the file is too large then exit
 720                 if ($curl_info["download_content_length"] > 1000000) {
 721                         return($url);
 722                 }
 723
 724                 // if it isn't a HTML file then exit
 725                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 726                         return($url);
 727                 }
 728
 729                 $stamp1 = microtime(true);
 730
 731                 $ch = curl_init();
 732                 curl_setopt($ch, CURLOPT_URL, $url);
 733                 curl_setopt($ch, CURLOPT_HEADER, 0);
 734                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 735                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 736                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 737                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 738
 739                 $body = curl_exec($ch);
 740                 curl_close($ch);
 741
 742                 $a->save_timestamp($stamp1, "network");
 743
 744                 if (trim($body) == "") {
 745                         return($url);
 746                 }
 747
 748                 // Check for redirect in meta elements
 749                 $doc = new DOMDocument();
 750                 @$doc->loadHTML($body);
 751
 752                 $xpath = new DomXPath($doc);
 753
 754                 $list = $xpath->query("//meta[@content]");
 755                 foreach ($list as $node) {
 756                         $attr = [];
 757                         if ($node->attributes->length) {
 758                                 foreach ($node->attributes as $attribute) {
 759                                         $attr[$attribute->name] = $attribute->value;
 760                                 }
 761                         }
 762
 763                         if (@$attr["http-equiv"] == 'refresh') {
 764                                 $path = $attr["content"];
 765                                 $pathinfo = explode(";", $path);
 766                                 foreach ($pathinfo as $value) {
 767                                         if (substr(strtolower($value), 0, 4) == "url=") {
 768                                                 return(self::finalUrl(substr($value, 4), ++$depth));
 769                                         }
 770                                 }
 771                         }
 772                 }
 773
 774                 return $url;
 775         }
 776
 777         /**
 778          * @brief Find the matching part between two url
 779          *
 780          * @param string $url1
 781          * @param string $url2
 782          * @return string The matching part
 783          */
 784         public static function getUrlMatch($url1, $url2)
 785         {
 786                 if (($url1 == "") || ($url2 == "")) {
 787                         return "";
 788                 }
 789
 790                 $url1 = normalise_link($url1);
 791                 $url2 = normalise_link($url2);
 792
 793                 $parts1 = parse_url($url1);
 794                 $parts2 = parse_url($url2);
 795
 796                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 797                         return "";
 798                 }
 799
 800                 if ($parts1["scheme"] != $parts2["scheme"]) {
 801                         return "";
 802                 }
 803
 804                 if ($parts1["host"] != $parts2["host"]) {
 805                         return "";
 806                 }
 807
 808                 if ($parts1["port"] != $parts2["port"]) {
 809                         return "";
 810                 }
 811
 812                 $match = $parts1["scheme"]."://".$parts1["host"];
 813
 814                 if ($parts1["port"]) {
 815                         $match .= ":".$parts1["port"];
 816                 }
 817
 818                 $pathparts1 = explode("/", $parts1["path"]);
 819                 $pathparts2 = explode("/", $parts2["path"]);
 820
 821                 $i = 0;
 822                 $path = "";
 823                 do {
 824                         $path1 = $pathparts1[$i];
 825                         $path2 = $pathparts2[$i];
 826
 827                         if ($path1 == $path2) {
 828                                 $path .= $path1."/";
 829                         }
 830                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 831
 832                 $match .= $path;
 833
 834                 return normalise_link($match);
 835         }
 836
 837         /**
 838          * @brief Glue url parts together
 839          *
 840          * @param array $parsed URL parts
 841          *
 842          * @return string The glued URL
 843          */
 844         public static function unparseURL($parsed)
 845         {
 846                 $get = function ($key) use ($parsed) {
 847                         return isset($parsed[$key]) ? $parsed[$key] : null;
 848                 };
 849
 850                 $pass      = $get('pass');
 851                 $user      = $get('user');
 852                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 853                 $port      = $get('port');
 854                 $scheme    = $get('scheme');
 855                 $query     = $get('query');
 856                 $fragment  = $get('fragment');
 857                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 858                                                 $get('host') .
 859                                                 ($port ? ":$port" : '');
 860
 861                 return  (strlen($scheme) ? $scheme.":" : '') .
 862                         (strlen($authority) ? "//".$authority : '') .
 863                         $get('path') .
 864                         (strlen($query) ? "?".$query : '') .
 865                         (strlen($fragment) ? "#".$fragment : '');
 866         }
 867 }