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
  16 require_once 'library/slinky.php';
  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('z_fetch_url: 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('fetch_url 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('fetch_url ' . $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::zFetchURL($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'] = $curl_info;
 251                         logger('z_fetch_url: error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 252                         logger('z_fetch_url: 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                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 374                         $newurl = trim(array_pop($matches));
 375
 376                         if (strpos($newurl, '/') === 0) {
 377                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 378                         }
 379
 380                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 381                                 $redirects++;
 382                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 383                                 return self::post($newurl, $params, $headers, $redirects, $timeout);
 384                         }
 385                 }
 386
 387                 $a->set_curl_code($http_code);
 388
 389                 $body = substr($s, strlen($header));
 390
 391                 $a->set_curl_headers($header);
 392
 393                 curl_close($ch);
 394
 395                 $a->save_timestamp($stamp1, 'network');
 396
 397                 logger('post_url: end ' . $url, LOGGER_DATA);
 398
 399                 return $body;
 400         }
 401
 402         /**
 403          * @brief Check URL to see if it's real
 404          *
 405          * Take a URL from the wild, prepend http:// if necessary
 406          * and check DNS to see if it's real (or check if is a valid IP address)
 407          *
 408          * @param string $url The URL to be validated
 409          * @return string|boolean The actual working URL, false else
 410          */
 411         public static function isUrlValid($url)
 412         {
 413                 if (Config::get('system', 'disable_url_validation')) {
 414                         return $url;
 415                 }
 416
 417                 // no naked subdomains (allow localhost for tests)
 418                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 419                         return false;
 420                 }
 421
 422                 if (substr($url, 0, 4) != 'http') {
 423                         $url = 'http://' . $url;
 424                 }
 425
 426                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 427                 $h = @parse_url($url);
 428
 429                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 430                         return $url;
 431                 }
 432
 433                 return false;
 434         }
 435
 436         /**
 437          * @brief Checks that email is an actual resolvable internet address
 438          *
 439          * @param string $addr The email address
 440          * @return boolean True if it's a valid email address, false if it's not
 441          */
 442         public static function isEmailDomainValid($addr)
 443         {
 444                 if (Config::get('system', 'disable_email_validation')) {
 445                         return true;
 446                 }
 447
 448                 if (! strpos($addr, '@')) {
 449                         return false;
 450                 }
 451
 452                 $h = substr($addr, strpos($addr, '@') + 1);
 453
 454                 if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 455                         return true;
 456                 }
 457                 return false;
 458         }
 459
 460         /**
 461          * @brief Check if URL is allowed
 462          *
 463          * Check $url against our list of allowed sites,
 464          * wildcards allowed. If allowed_sites is unset return true;
 465          *
 466          * @param string $url URL which get tested
 467          * @return boolean True if url is allowed otherwise return false
 468          */
 469         public static function isUrlAllowed($url)
 470         {
 471                 $h = @parse_url($url);
 472
 473                 if (! $h) {
 474                         return false;
 475                 }
 476
 477                 $str_allowed = Config::get('system', 'allowed_sites');
 478                 if (! $str_allowed) {
 479                         return true;
 480                 }
 481
 482                 $found = false;
 483
 484                 $host = strtolower($h['host']);
 485
 486                 // always allow our own site
 487                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 488                         return true;
 489                 }
 490
 491                 $fnmatch = function_exists('fnmatch');
 492                 $allowed = explode(',', $str_allowed);
 493
 494                 if (count($allowed)) {
 495                         foreach ($allowed as $a) {
 496                                 $pat = strtolower(trim($a));
 497                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 498                                         $found = true;
 499                                         break;
 500                                 }
 501                         }
 502                 }
 503                 return $found;
 504         }
 505
 506         /**
 507          * Checks if the provided url domain is on the domain blocklist.
 508          * Returns true if it is or malformed URL, false if not.
 509          *
 510          * @param string $url The url to check the domain from
 511          *
 512          * @return boolean
 513          */
 514         public static function isUrlBlocked($url)
 515         {
 516                 $h = @parse_url($url);
 517
 518                 if (! $h) {
 519                         return true;
 520                 }
 521
 522                 $domain_blocklist = Config::get('system', 'blocklist', []);
 523                 if (! $domain_blocklist) {
 524                         return false;
 525                 }
 526
 527                 $host = strtolower($h['host']);
 528
 529                 foreach ($domain_blocklist as $domain_block) {
 530                         if (strtolower($domain_block['domain']) == $host) {
 531                                 return true;
 532                         }
 533                 }
 534
 535                 return false;
 536         }
 537
 538         /**
 539          * @brief Check if email address is allowed to register here.
 540          *
 541          * Compare against our list (wildcards allowed).
 542          *
 543          * @param  string $email email address
 544          * @return boolean False if not allowed, true if allowed
 545          *    or if allowed list is not configured
 546          */
 547         public static function isEmailDomainAllowed($email)
 548         {
 549                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 550                 if (!$domain) {
 551                         return false;
 552                 }
 553
 554                 $str_allowed = Config::get('system', 'allowed_email', '');
 555                 if (!x($str_allowed)) {
 556                         return true;
 557                 }
 558
 559                 $allowed = explode(',', $str_allowed);
 560
 561                 return self::isDomainAllowed($domain, $allowed);
 562         }
 563
 564         /**
 565          * Checks for the existence of a domain in a domain list
 566          *
 567          * @brief Checks for the existence of a domain in a domain list
 568          * @param string $domain
 569          * @param array  $domain_list
 570          * @return boolean
 571          */
 572         public static function isDomainAllowed($domain, array $domain_list)
 573         {
 574                 $found = false;
 575
 576                 foreach ($domain_list as $item) {
 577                         $pat = strtolower(trim($item));
 578                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 579                                 $found = true;
 580                                 break;
 581                         }
 582                 }
 583
 584                 return $found;
 585         }
 586
 587         public static function lookupAvatarByEmail($email)
 588         {
 589                 $avatar['size'] = 175;
 590                 $avatar['email'] = $email;
 591                 $avatar['url'] = '';
 592                 $avatar['success'] = false;
 593
 594                 Addon::callHooks('avatar_lookup', $avatar);
 595
 596                 if (! $avatar['success']) {
 597                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 598                 }
 599
 600                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 601                 return $avatar['url'];
 602         }
 603
 604         /**
 605          * @brief Remove Google Analytics and other tracking platforms params from URL
 606          *
 607          * @param string $url Any user-submitted URL that may contain tracking params
 608          * @return string The same URL stripped of tracking parameters
 609          */
 610         public static function stripTrackingQueryParams($url)
 611         {
 612                 $urldata = parse_url($url);
 613                 if (is_string($urldata["query"])) {
 614                         $query = $urldata["query"];
 615                         parse_str($query, $querydata);
 616
 617                         if (is_array($querydata)) {
 618                                 foreach ($querydata as $param => $value) {
 619                                         if (in_array(
 620                                                 $param,
 621                                                 [
 622                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 623                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 624                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 625                                                         "awesm", "wtrid",
 626                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 627                                                 )
 628                                         ) {
 629                                                 $pair = $param . "=" . urlencode($value);
 630                                                 $url = str_replace($pair, "", $url);
 631
 632                                                 // Second try: if the url isn't encoded completely
 633                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 634                                                 $url = str_replace($pair, "", $url);
 635
 636                                                 // Third try: Maybey the url isn't encoded at all
 637                                                 $pair = $param . "=" . $value;
 638                                                 $url = str_replace($pair, "", $url);
 639
 640                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 641                                         }
 642                                 }
 643                         }
 644
 645                         if (substr($url, -1, 1) == "?") {
 646                                 $url = substr($url, 0, -1);
 647                         }
 648                 }
 649
 650                 return $url;
 651         }
 652
 653         /**
 654          * @brief Returns the original URL of the provided URL
 655          *
 656          * This function strips tracking query params and follows redirections, either
 657          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 658          *
 659          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 660          *
 661          * @see ParseUrl::getSiteinfo
 662          *
 663          * @param string $url       A user-submitted URL
 664          * @param int    $depth     The current redirection recursion level (internal)
 665          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 666          * @return string A canonical URL
 667          */
 668         public static function finalUrl($url, $depth = 1, $fetchbody = false)
 669         {
 670                 $a = get_app();
 671
 672                 $url = self::stripTrackingQueryParams($url);
 673
 674                 if ($depth > 10) {
 675                         return($url);
 676                 }
 677
 678                 $url = trim($url, "'");
 679
 680                 $stamp1 = microtime(true);
 681
 682                 $ch = curl_init();
 683                 curl_setopt($ch, CURLOPT_URL, $url);
 684                 curl_setopt($ch, CURLOPT_HEADER, 1);
 685                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 686                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 687                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 688                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 689
 690                 curl_exec($ch);
 691                 $curl_info = @curl_getinfo($ch);
 692                 $http_code = $curl_info['http_code'];
 693                 curl_close($ch);
 694
 695                 $a->save_timestamp($stamp1, "network");
 696
 697                 if ($http_code == 0) {
 698                         return($url);
 699                 }
 700
 701                 if ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 702                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 703                 ) {
 704                         if ($curl_info['redirect_url'] != "") {
 705                                 return(self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody));
 706                         } else {
 707                                 return(self::finalUrl($curl_info['location'], ++$depth, $fetchbody));
 708                         }
 709                 }
 710
 711                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 712                 if (!$fetchbody) {
 713                         return(self::finalUrl($url, ++$depth, true));
 714                 }
 715
 716                 // if the file is too large then exit
 717                 if ($curl_info["download_content_length"] > 1000000) {
 718                         return($url);
 719                 }
 720
 721                 // if it isn't a HTML file then exit
 722                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 723                         return($url);
 724                 }
 725
 726                 $stamp1 = microtime(true);
 727
 728                 $ch = curl_init();
 729                 curl_setopt($ch, CURLOPT_URL, $url);
 730                 curl_setopt($ch, CURLOPT_HEADER, 0);
 731                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 732                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 733                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 734                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 735
 736                 $body = curl_exec($ch);
 737                 curl_close($ch);
 738
 739                 $a->save_timestamp($stamp1, "network");
 740
 741                 if (trim($body) == "") {
 742                         return($url);
 743                 }
 744
 745                 // Check for redirect in meta elements
 746                 $doc = new DOMDocument();
 747                 @$doc->loadHTML($body);
 748
 749                 $xpath = new DomXPath($doc);
 750
 751                 $list = $xpath->query("//meta[@content]");
 752                 foreach ($list as $node) {
 753                         $attr = [];
 754                         if ($node->attributes->length) {
 755                                 foreach ($node->attributes as $attribute) {
 756                                         $attr[$attribute->name] = $attribute->value;
 757                                 }
 758                         }
 759
 760                         if (@$attr["http-equiv"] == 'refresh') {
 761                                 $path = $attr["content"];
 762                                 $pathinfo = explode(";", $path);
 763                                 foreach ($pathinfo as $value) {
 764                                         if (substr(strtolower($value), 0, 4) == "url=") {
 765                                                 return(self::finalUrl(substr($value, 4), ++$depth));
 766                                         }
 767                                 }
 768                         }
 769                 }
 770
 771                 return $url;
 772         }
 773
 774         public static function shortenUrl($url)
 775         {
 776                 $slinky = new Slinky($url);
 777                 $yourls_url = Config::get('yourls', 'url1');
 778                 if ($yourls_url) {
 779                         $yourls_username = Config::get('yourls', 'username1');
 780                         $yourls_password = Config::get('yourls', 'password1');
 781                         $yourls_ssl = Config::get('yourls', 'ssl1');
 782                         $yourls = new Slinky_YourLS();
 783                         $yourls->set('username', $yourls_username);
 784                         $yourls->set('password', $yourls_password);
 785                         $yourls->set('ssl', $yourls_ssl);
 786                         $yourls->set('yourls-url', $yourls_url);
 787                         $slinky->set_cascade([$yourls, new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 788                 } else {
 789                         // setup a cascade of shortening services
 790                         // try to get a short link from these services
 791                         // in the order ur1.ca, tinyurl
 792                         $slinky->set_cascade([new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 793                 }
 794                 return $slinky->short();
 795         }
 796
 797         /**
 798          * @brief Find the matching part between two url
 799          *
 800          * @param string $url1
 801          * @param string $url2
 802          * @return string The matching part
 803          */
 804         public static function getUrlMatch($url1, $url2)
 805         {
 806                 if (($url1 == "") || ($url2 == "")) {
 807                         return "";
 808                 }
 809
 810                 $url1 = normalise_link($url1);
 811                 $url2 = normalise_link($url2);
 812
 813                 $parts1 = parse_url($url1);
 814                 $parts2 = parse_url($url2);
 815
 816                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
 817                         return "";
 818                 }
 819
 820                 if ($parts1["scheme"] != $parts2["scheme"]) {
 821                         return "";
 822                 }
 823
 824                 if ($parts1["host"] != $parts2["host"]) {
 825                         return "";
 826                 }
 827
 828                 if ($parts1["port"] != $parts2["port"]) {
 829                         return "";
 830                 }
 831
 832                 $match = $parts1["scheme"]."://".$parts1["host"];
 833
 834                 if ($parts1["port"]) {
 835                         $match .= ":".$parts1["port"];
 836                 }
 837
 838                 $pathparts1 = explode("/", $parts1["path"]);
 839                 $pathparts2 = explode("/", $parts2["path"]);
 840
 841                 $i = 0;
 842                 $path = "";
 843                 do {
 844                         $path1 = $pathparts1[$i];
 845                         $path2 = $pathparts2[$i];
 846
 847                         if ($path1 == $path2) {
 848                                 $path .= $path1."/";
 849                         }
 850                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
 851
 852                 $match .= $path;
 853
 854                 return normalise_link($match);
 855         }
 856
 857         /**
 858          * @brief Glue url parts together
 859          *
 860          * @param array $parsed URL parts
 861          *
 862          * @return string The glued URL
 863          */
 864         public static function unparseURL($parsed)
 865         {
 866                 $get = function ($key) use ($parsed) {
 867                         return isset($parsed[$key]) ? $parsed[$key] : null;
 868                 };
 869
 870                 $pass      = $get('pass');
 871                 $user      = $get('user');
 872                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
 873                 $port      = $get('port');
 874                 $scheme    = $get('scheme');
 875                 $query     = $get('query');
 876                 $fragment  = $get('fragment');
 877                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
 878                                                 $get('host') .
 879                                                 ($port ? ":$port" : '');
 880
 881                 return  (strlen($scheme) ? $scheme.":" : '') .
 882                         (strlen($authority) ? "//".$authority : '') .
 883                         $get('path') .
 884                         (strlen($query) ? "?".$query : '') .
 885                         (strlen($fragment) ? "#".$fragment : '');
 886         }
 887 }