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                         $old_location_query = @parse_url($url, PHP_URL_QUERY);
 233
 234                         if ($old_location_query != '') {
 235                                 $newurl .= '?' . $old_location_query;
 236                         }
 237
 238                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 239                                 $redirects++;
 240                                 @curl_close($ch);
 241                                 return self::curl($newurl, $binary, $redirects, $opts);
 242                         }
 243                 }
 244
 245                 $a->set_curl_code($http_code);
 246                 $a->set_curl_content_type($curl_info['content_type']);
 247
 248                 $rc = intval($http_code);
 249                 $ret['return_code'] = $rc;
 250                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 251                 $ret['redirect_url'] = $url;
 252
 253                 if (!$ret['success']) {
 254                         $ret['error'] = curl_error($ch);
 255                         $ret['debug'] = $curl_info;
 256                         logger('error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 257                         logger('debug: '.print_r($curl_info, true), LOGGER_DATA);
 258                 }
 259
 260                 $ret['body'] = substr($s, strlen($header));
 261                 $ret['header'] = $header;
 262
 263                 if (x($opts, 'debug')) {
 264                         $ret['debug'] = $curl_info;
 265                 }
 266
 267                 @curl_close($ch);
 268
 269                 $a->save_timestamp($stamp1, 'network');
 270
 271                 return($ret);
 272         }
 273
 274         /**
 275          * @brief Send POST request to $url
 276          *
 277          * @param string  $url       URL to post
 278          * @param mixed   $params    array of POST variables
 279          * @param string  $headers   HTTP headers
 280          * @param integer $redirects Recursion counter for internal use - default = 0
 281          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 282          *
 283          * @return string The content
 284          */
 285         public static function post($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 286         {
 287                 $stamp1 = microtime(true);
 288
 289                 if (self::isUrlBlocked($url)) {
 290                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 291                         return false;
 292                 }
 293
 294                 $a = get_app();
 295                 $ch = curl_init($url);
 296
 297                 if (($redirects > 8) || (!$ch)) {
 298                         return false;
 299                 }
 300
 301                 logger('post_url: start ' . $url, LOGGER_DATA);
 302
 303                 curl_setopt($ch, CURLOPT_HEADER, true);
 304                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 305                 curl_setopt($ch, CURLOPT_POST, 1);
 306                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 307                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 308
 309                 if (Config::get('system', 'ipv4_resolve', false)) {
 310                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 311                 }
 312
 313                 if (intval($timeout)) {
 314                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 315                 } else {
 316                         $curl_time = Config::get('system', 'curl_timeout', 60);
 317                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 318                 }
 319
 320                 if (defined('LIGHTTPD')) {
 321                         if (!is_array($headers)) {
 322                                 $headers = ['Expect:'];
 323                         } else {
 324                                 if (!in_array('Expect:', $headers)) {
 325                                         array_push($headers, 'Expect:');
 326                                 }
 327                         }
 328                 }
 329
 330                 if ($headers) {
 331                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 332                 }
 333
 334                 $check_cert = Config::get('system', 'verifyssl');
 335                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 336
 337                 if ($check_cert) {
 338                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 339                 }
 340
 341                 $proxy = Config::get('system', 'proxy');
 342
 343                 if (strlen($proxy)) {
 344                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 345                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 346                         $proxyuser = Config::get('system', 'proxyuser');
 347                         if (strlen($proxyuser)) {
 348                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 349                         }
 350                 }
 351
 352                 $a->set_curl_code(0);
 353
 354                 // don't let curl abort the entire application
 355                 // if it throws any errors.
 356
 357                 $s = @curl_exec($ch);
 358
 359                 $base = $s;
 360                 $curl_info = curl_getinfo($ch);
 361                 $http_code = $curl_info['http_code'];
 362
 363                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 364
 365                 $header = '';
 366
 367                 // Pull out multiple headers, e.g. proxy and continuation headers
 368                 // allow for HTTP/2.x without fixing code
 369
 370                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 371                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 372                         $header .= $chunk;
 373                         $base = substr($base, strlen($chunk));
 374                 }
 375
 376                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 377                         $matches = [];
 378                         $new_location_info = @parse_url($curl_info['redirect_url']);
 379                         $old_location_info = @parse_url($curl_info['url']);
 380
 381                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 382                         $newurl = trim(array_pop($matches));
 383
 384                         if (strpos($newurl, '/') === 0) {
 385                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 386                         }
 387
 388                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 389                                 $redirects++;
 390                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 391                                 return self::post($newurl, $params, $headers, $redirects, $timeout);
 392                         }
 393                 }
 394
 395                 $a->set_curl_code($http_code);
 396
 397                 $body = substr($s, strlen($header));
 398
 399                 $a->set_curl_headers($header);
 400
 401                 curl_close($ch);
 402
 403                 $a->save_timestamp($stamp1, 'network');
 404
 405                 logger('post_url: end ' . $url, LOGGER_DATA);
 406
 407                 return $body;
 408         }
 409
 410         /**
 411          * @brief Check URL to see if it's real
 412          *
 413          * Take a URL from the wild, prepend http:// if necessary
 414          * and check DNS to see if it's real (or check if is a valid IP address)
 415          *
 416          * @param string $url The URL to be validated
 417          * @return string|boolean The actual working URL, false else
 418          */
 419         public static function isUrlValid($url)
 420         {
 421                 if (Config::get('system', 'disable_url_validation')) {
 422                         return $url;
 423                 }
 424
 425                 // no naked subdomains (allow localhost for tests)
 426                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 427                         return false;
 428                 }
 429
 430                 if (substr($url, 0, 4) != 'http') {
 431                         $url = 'http://' . $url;
 432                 }
 433
 434                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 435                 $h = @parse_url($url);
 436
 437                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 438                         return $url;
 439                 }
 440
 441                 return false;
 442         }
 443
 444         /**
 445          * @brief Checks that email is an actual resolvable internet address
 446          *
 447          * @param string $addr The email address
 448          * @return boolean True if it's a valid email address, false if it's not
 449          */
 450         public static function isEmailDomainValid($addr)
 451         {
 452                 if (Config::get('system', 'disable_email_validation')) {
 453                         return true;
 454                 }
 455
 456                 if (! strpos($addr, '@')) {
 457                         return false;
 458                 }
 459
 460                 $h = substr($addr, strpos($addr, '@') + 1);
 461
 462                 if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 463                         return true;
 464                 }
 465                 return false;
 466         }
 467
 468         /**
 469          * @brief Check if URL is allowed
 470          *
 471          * Check $url against our list of allowed sites,
 472          * wildcards allowed. If allowed_sites is unset return true;
 473          *
 474          * @param string $url URL which get tested
 475          * @return boolean True if url is allowed otherwise return false
 476          */
 477         public static function isUrlAllowed($url)
 478         {
 479                 $h = @parse_url($url);
 480
 481                 if (! $h) {
 482                         return false;
 483                 }
 484
 485                 $str_allowed = Config::get('system', 'allowed_sites');
 486                 if (! $str_allowed) {
 487                         return true;
 488                 }
 489
 490                 $found = false;
 491
 492                 $host = strtolower($h['host']);
 493
 494                 // always allow our own site
 495                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 496                         return true;
 497                 }
 498
 499                 $fnmatch = function_exists('fnmatch');
 500                 $allowed = explode(',', $str_allowed);
 501
 502                 if (count($allowed)) {
 503                         foreach ($allowed as $a) {
 504                                 $pat = strtolower(trim($a));
 505                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 506                                         $found = true;
 507                                         break;
 508                                 }
 509                         }
 510                 }
 511                 return $found;
 512         }
 513
 514         /**
 515          * Checks if the provided url domain is on the domain blocklist.
 516          * Returns true if it is or malformed URL, false if not.
 517          *
 518          * @param string $url The url to check the domain from
 519          *
 520          * @return boolean
 521          */
 522         public static function isUrlBlocked($url)
 523         {
 524                 $h = @parse_url($url);
 525
 526                 if (! $h) {
 527                         return true;
 528                 }
 529
 530                 $domain_blocklist = Config::get('system', 'blocklist', []);
 531                 if (! $domain_blocklist) {
 532                         return false;
 533                 }
 534
 535                 $host = strtolower($h['host']);
 536
 537                 foreach ($domain_blocklist as $domain_block) {
 538                         if (strtolower($domain_block['domain']) == $host) {
 539                                 return true;
 540                         }
 541                 }
 542
 543                 return false;
 544         }
 545
 546         /**
 547          * @brief Check if email address is allowed to register here.
 548          *
 549          * Compare against our list (wildcards allowed).
 550          *
 551          * @param  string $email email address
 552          * @return boolean False if not allowed, true if allowed
 553          *    or if allowed list is not configured
 554          */
 555         public static function isEmailDomainAllowed($email)
 556         {
 557                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 558                 if (!$domain) {
 559                         return false;
 560                 }
 561
 562                 $str_allowed = Config::get('system', 'allowed_email', '');
 563                 if (!x($str_allowed)) {
 564                         return true;
 565                 }
 566
 567                 $allowed = explode(',', $str_allowed);
 568
 569                 return self::isDomainAllowed($domain, $allowed);
 570         }
 571
 572         /**
 573          * Checks for the existence of a domain in a domain list
 574          *
 575          * @brief Checks for the existence of a domain in a domain list
 576          * @param string $domain
 577          * @param array  $domain_list
 578          * @return boolean
 579          */
 580         public static function isDomainAllowed($domain, array $domain_list)
 581         {
 582                 $found = false;
 583
 584                 foreach ($domain_list as $item) {
 585                         $pat = strtolower(trim($item));
 586                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 587                                 $found = true;
 588                                 break;
 589                         }
 590                 }
 591
 592                 return $found;
 593         }
 594
 595         public static function lookupAvatarByEmail($email)
 596         {
 597                 $avatar['size'] = 175;
 598                 $avatar['email'] = $email;
 599                 $avatar['url'] = '';
 600                 $avatar['success'] = false;
 601
 602                 Addon::callHooks('avatar_lookup', $avatar);
 603
 604                 if (! $avatar['success']) {
 605                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 606                 }
 607
 608                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 609                 return $avatar['url'];
 610         }
 611
 612         /**
 613          * @brief Remove Google Analytics and other tracking platforms params from URL
 614          *
 615          * @param string $url Any user-submitted URL that may contain tracking params
 616          * @return string The same URL stripped of tracking parameters
 617          */
 618         public static function stripTrackingQueryParams($url)
 619         {
 620                 $urldata = parse_url($url);
 621                 if (is_string($urldata["query"])) {
 622                         $query = $urldata["query"];
 623                         parse_str($query, $querydata);
 624
 625                         if (is_array($querydata)) {
 626                                 foreach ($querydata as $param => $value) {
 627                                         if (in_array(
 628                                                 $param,
 629                                                 [
 630                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 631                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 632                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 633                                                         "awesm", "wtrid",
 634                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 635                                                 )
 636                                         ) {
 637                                                 $pair = $param . "=" . urlencode($value);
 638                                                 $url = str_replace($pair, "", $url);
 639
 640                                                 // Second try: if the url isn't encoded completely
 641                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 642                                                 $url = str_replace($pair, "", $url);
 643
 644                                                 // Third try: Maybey the url isn't encoded at all
 645                                                 $pair = $param . "=" . $value;
 646                                                 $url = str_replace($pair, "", $url);
 647
 648                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 649                                         }
 650                                 }
 651                         }
 652
 653                         if (substr($url, -1, 1) == "?") {
 654                                 $url = substr($url, 0, -1);
 655                         }
 656                 }
 657
 658                 return $url;
 659         }
 660
 661         /**
 662          * @brief Returns the original URL of the provided URL
 663          *
 664          * This function strips tracking query params and follows redirections, either
 665          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 666          *
 667          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 668          *
 669          * @see ParseUrl::getSiteinfo
 670          *
 671          * @param string $url       A user-submitted URL
 672          * @param int    $depth     The current redirection recursion level (internal)
 673          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 674          * @return string A canonical URL
 675          */
 676         public static function finalUrl($url, $depth = 1, $fetchbody = false)
 677         {
 678                 $a = get_app();
 679
 680                 $url = self::stripTrackingQueryParams($url);
 681
 682                 if ($depth > 10) {
 683                         return($url);
 684                 }
 685
 686                 $url = trim($url, "'");
 687
 688                 $stamp1 = microtime(true);
 689
 690                 $ch = curl_init();
 691                 curl_setopt($ch, CURLOPT_URL, $url);
 692                 curl_setopt($ch, CURLOPT_HEADER, 1);
 693                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 694                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 695                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 696                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 697
 698                 curl_exec($ch);
 699                 $curl_info = @curl_getinfo($ch);
 700                 $http_code = $curl_info['http_code'];
 701                 curl_close($ch);
 702
 703                 $a->save_timestamp($stamp1, "network");
 704
 705                 if ($http_code == 0) {
 706                         return($url);
 707                 }
 708
 709                 if ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 710                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 711                 ) {
 712                         if ($curl_info['redirect_url'] != "") {
 713                                 return(self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody));
 714                         } else {
 715                                 return(self::finalUrl($curl_info['location'], ++$depth, $fetchbody));
 716                         }
 717                 }
 718
 719                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 720                 if (!$fetchbody) {
 721                         return(self::finalUrl($url, ++$depth, true));
 722                 }
 723
 724                 // if the file is too large then exit
 725                 if ($curl_info["download_content_length"] > 1000000) {
 726                         return($url);
 727                 }
 728
 729                 // if it isn't a HTML file then exit
 730                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 731                         return($url);
 732                 }
 733
 734                 $stamp1 = microtime(true);
 735
 736                 $ch = curl_init();
 737                 curl_setopt($ch, CURLOPT_URL, $url);
 738                 curl_setopt($ch, CURLOPT_HEADER, 0);
 739                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 740                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 741                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 742                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 743
 744                 $body = curl_exec($ch);
 745                 curl_close($ch);
 746
 747                 $a->save_timestamp($stamp1, "network");
 748
 749                 if (trim($body) == "") {
 750                         return($url);
 751                 }
 752
 753                 // Check for redirect in meta elements
 754                 $doc = new DOMDocument();
 755                 @$doc->loadHTML($body);
 756
 757                 $xpath = new DomXPath($doc);
 758
 759                 $list = $xpath->query("//meta[@content]");
 760                 foreach ($list as $node) {
 761                         $attr = [];
 762                         if ($node->attributes->length) {
 763                                 foreach ($node->attributes as $attribute) {
 764                                         $attr[$attribute->name] = $attribute->value;
 765                                 }
 766                         }
 767
 768                         if (@$attr["http-equiv"] == 'refresh') {
 769                                 $path = $attr["content"];
 770                                 $pathinfo = explode(";", $path);
 771                                 foreach ($pathinfo as $value) {
 772                                         if (substr(strtolower($value), 0, 4) == "url=") {
 773                                                 return(self::finalUrl(substr($value, 4), ++$depth));
 774                                         }
 775                                 }
 776                         }
 777                 }
 778
 779                 return $url;
 780         }
 781
 782         /**
 783          * @brief Find the matching part between two url
 784          *
 785          * @param string $url1
 786          * @param string $url2
 787          * @return string The matching part
 788          */
 789         public static function getUrlMatch($url1, $url2)
 790         {
 791                 if (($url1 == "") || ($url2 == "")) {
 792                         return "";
 793                 }
 794
 795                 $url1 = normalise_link($url1);
 796                 $url2 = normalise_link($url2);
 797
 798                 $parts1 = parse_url($url1);
 799                 $parts2 = parse_url($url2);
 800
 801                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 802                         return "";
 803                 }
 804
 805                 if ($parts1["scheme"] != $parts2["scheme"]) {
 806                         return "";
 807                 }
 808
 809                 if ($parts1["host"] != $parts2["host"]) {
 810                         return "";
 811                 }
 812
 813                 if ($parts1["port"] != $parts2["port"]) {
 814                         return "";
 815                 }
 816
 817                 $match = $parts1["scheme"]."://".$parts1["host"];
 818
 819                 if ($parts1["port"]) {
 820                         $match .= ":".$parts1["port"];
 821                 }
 822
 823                 $pathparts1 = explode("/", $parts1["path"]);
 824                 $pathparts2 = explode("/", $parts2["path"]);
 825
 826                 $i = 0;
 827                 $path = "";
 828                 do {
 829                         $path1 = $pathparts1[$i];
 830                         $path2 = $pathparts2[$i];
 831
 832                         if ($path1 == $path2) {
 833                                 $path .= $path1."/";
 834                         }
 835                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 836
 837                 $match .= $path;
 838
 839                 return normalise_link($match);
 840         }
 841
 842         /**
 843          * @brief Glue url parts together
 844          *
 845          * @param array $parsed URL parts
 846          *
 847          * @return string The glued URL
 848          */
 849         public static function unparseURL($parsed)
 850         {
 851                 $get = function ($key) use ($parsed) {
 852                         return isset($parsed[$key]) ? $parsed[$key] : null;
 853                 };
 854
 855                 $pass      = $get('pass');
 856                 $user      = $get('user');
 857                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 858                 $port      = $get('port');
 859                 $scheme    = $get('scheme');
 860                 $query     = $get('query');
 861                 $fragment  = $get('fragment');
 862                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 863                                                 $get('host') .
 864                                                 ($port ? ":$port" : '');
 865
 866                 return  (strlen($scheme) ? $scheme.":" : '') .
 867                         (strlen($authority) ? "//".$authority : '') .
 868                         $get('path') .
 869                         (strlen($query) ? "?".$query : '') .
 870                         (strlen($fragment) ? "#".$fragment : '');
 871         }
 872 }