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\Network;
  15 use Friendica\Util\XML;
  16
  17 require_once 'library/slinky.php';
  18
  19 class Network
  20 {
  21         /**
  22          * @brief Curl wrapper
  23          *
  24          * If binary flag is true, return binary results.
  25          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
  26          * to preserve cookies from one request to the next.
  27          *
  28          * @param string  $url            URL to fetch
  29          * @param boolean $binary         default false
  30          *                                TRUE if asked to return binary results (file download)
  31          * @param integer $redirects      The recursion counter for internal use - default 0
  32          * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
  33          * @param string  $accept_content supply Accept: header with 'accept_content' as the value
  34          * @param string  $cookiejar      Path to cookie jar file
  35          *
  36          * @return string The fetched content
  37          */
  38         public static function fetchURL($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
  39         {
  40                 $ret = self::zFetchURL(
  41                         $url,
  42                         $binary,
  43                         $redirects,
  44                         ['timeout'=>$timeout,
  45                         'accept_content'=>$accept_content,
  46                         'cookiejar'=>$cookiejar
  47                         ]
  48                 );
  49
  50                 return($ret['body']);
  51         }
  52
  53         /**
  54          * @brief fetches an URL.
  55          *
  56          * @param string  $url       URL to fetch
  57          * @param boolean $binary    default false
  58          *                           TRUE if asked to return binary results (file download)
  59          * @param int     $redirects The recursion counter for internal use - default 0
  60          * @param array   $opts      (optional parameters) assoziative array with:
  61          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  62          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  63          *                           'http_auth' => username:password
  64          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  65          *                           'nobody' => only return the header
  66          *                           'cookiejar' => path to cookie jar file
  67          *
  68          * @return array an assoziative array with:
  69          *    int 'return_code' => HTTP return code or 0 if timeout or failure
  70          *    boolean 'success' => boolean true (if HTTP 2xx result) or false
  71          *    string 'redirect_url' => in case of redirect, content was finally retrieved from this URL
  72          *    string 'header' => HTTP headers
  73          *    string 'body' => fetched content
  74          */
  75         public static function zFetchURL($url, $binary = false, &$redirects = 0, $opts = [])
  76         {
  77                 $ret = ['return_code' => 0, 'success' => false, 'header' => '', 'info' => '', 'body' => ''];
  78
  79                 $stamp1 = microtime(true);
  80
  81                 $a = get_app();
  82
  83                 if (self::blockedURL($url)) {
  84                         logger('z_fetch_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
  85                         return $ret;
  86                 }
  87
  88                 $ch = @curl_init($url);
  89
  90                 if (($redirects > 8) || (!$ch)) {
  91                         return $ret;
  92                 }
  93
  94                 @curl_setopt($ch, CURLOPT_HEADER, true);
  95
  96                 if (x($opts, "cookiejar")) {
  97                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
  98                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
  99                 }
 100
 101                 // These settings aren't needed. We're following the location already.
 102                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 103                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 104
 105                 if (x($opts, 'accept_content')) {
 106                         curl_setopt(
 107                                 $ch,
 108                                 CURLOPT_HTTPHEADER,
 109                                 ['Accept: ' . $opts['accept_content']]
 110                         );
 111                 }
 112
 113                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 114                 @curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 115
 116                 $range = intval(Config::get('system', 'curl_range_bytes', 0));
 117
 118                 if ($range > 0) {
 119                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 120                 }
 121
 122                 // Without this setting it seems as if some webservers send compressed content
 123                 // This seems to confuse curl so that it shows this uncompressed.
 124                 /// @todo  We could possibly set this value to "gzip" or something similar
 125                 curl_setopt($ch, CURLOPT_ENCODING, '');
 126
 127                 if (x($opts, 'headers')) {
 128                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 129                 }
 130
 131                 if (x($opts, 'nobody')) {
 132                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 133                 }
 134
 135                 if (x($opts, 'timeout')) {
 136                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 137                 } else {
 138                         $curl_time = Config::get('system', 'curl_timeout', 60);
 139                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 140                 }
 141
 142                 // by default we will allow self-signed certs
 143                 // but you can override this
 144
 145                 $check_cert = Config::get('system', 'verifyssl');
 146                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 147
 148                 if ($check_cert) {
 149                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 150                 }
 151
 152                 $proxy = Config::get('system', 'proxy');
 153
 154                 if (strlen($proxy)) {
 155                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 156                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 157                         $proxyuser = @Config::get('system', 'proxyuser');
 158
 159                         if (strlen($proxyuser)) {
 160                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 161                         }
 162                 }
 163
 164                 if (Config::get('system', 'ipv4_resolve', false)) {
 165                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 166                 }
 167
 168                 if ($binary) {
 169                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 170                 }
 171
 172                 $a->set_curl_code(0);
 173
 174                 // don't let curl abort the entire application
 175                 // if it throws any errors.
 176
 177                 $s = @curl_exec($ch);
 178                 $curl_info = @curl_getinfo($ch);
 179
 180                 // Special treatment for HTTP Code 416
 181                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 182                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 183                         @curl_setopt($ch, CURLOPT_RANGE, '');
 184                         $s = @curl_exec($ch);
 185                         $curl_info = @curl_getinfo($ch);
 186                 }
 187
 188                 if (curl_errno($ch) !== CURLE_OK) {
 189                         logger('fetch_url error fetching ' . $url . ': ' . curl_error($ch), LOGGER_NORMAL);
 190                 }
 191
 192                 $ret['errno'] = curl_errno($ch);
 193
 194                 $base = $s;
 195                 $ret['info'] = $curl_info;
 196
 197                 $http_code = $curl_info['http_code'];
 198
 199                 logger('fetch_url ' . $url . ': ' . $http_code . " " . $s, LOGGER_DATA);
 200                 $header = '';
 201
 202                 // Pull out multiple headers, e.g. proxy and continuation headers
 203                 // allow for HTTP/2.x without fixing code
 204
 205                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 206                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 207                         $header .= $chunk;
 208                         $base = substr($base, strlen($chunk));
 209                 }
 210
 211                 $a->set_curl_code($http_code);
 212                 $a->set_curl_content_type($curl_info['content_type']);
 213                 $a->set_curl_headers($header);
 214
 215                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 216                         $new_location_info = @parse_url($curl_info['redirect_url']);
 217                         $old_location_info = @parse_url($curl_info['url']);
 218
 219                         $newurl = $curl_info['redirect_url'];
 220
 221                         if (($new_location_info['path'] == '') && ( $new_location_info['host'] != '')) {
 222                                 $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
 223                         }
 224
 225                         $matches = [];
 226
 227                         if (preg_match('/(Location:|URI:)(.*?)\n/i', $header, $matches)) {
 228                                 $newurl = trim(array_pop($matches));
 229                         }
 230                         if (strpos($newurl, '/') === 0) {
 231                                 $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
 232                         }
 233
 234                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 235                                 $redirects++;
 236                                 @curl_close($ch);
 237                                 return self::zFetchURL($newurl, $binary, $redirects, $opts);
 238                         }
 239                 }
 240
 241                 $a->set_curl_code($http_code);
 242                 $a->set_curl_content_type($curl_info['content_type']);
 243
 244                 $rc = intval($http_code);
 245                 $ret['return_code'] = $rc;
 246                 $ret['success'] = (($rc >= 200 && $rc <= 299) ? true : false);
 247                 $ret['redirect_url'] = $url;
 248
 249                 if (!$ret['success']) {
 250                         $ret['error'] = curl_error($ch);
 251                         $ret['debug'] = $curl_info;
 252                         logger('z_fetch_url: error: '.$url.': '.$ret['return_code'].' - '.$ret['error'], LOGGER_DEBUG);
 253                         logger('z_fetch_url: debug: '.print_r($curl_info, true), LOGGER_DATA);
 254                 }
 255
 256                 $ret['body'] = substr($s, strlen($header));
 257                 $ret['header'] = $header;
 258
 259                 if (x($opts, 'debug')) {
 260                         $ret['debug'] = $curl_info;
 261                 }
 262
 263                 @curl_close($ch);
 264
 265                 $a->save_timestamp($stamp1, 'network');
 266
 267                 return($ret);
 268         }
 269
 270         /**
 271          * @brief Send POST request to $url
 272          *
 273          * @param string  $url       URL to post
 274          * @param mixed   $params    array of POST variables
 275          * @param string  $headers   HTTP headers
 276          * @param integer $redirects Recursion counter for internal use - default = 0
 277          * @param integer $timeout   The timeout in seconds, default system config value or 60 seconds
 278          *
 279          * @return string The content
 280          */
 281         public static function postURL($url, $params, $headers = null, &$redirects = 0, $timeout = 0)
 282         {
 283                 $stamp1 = microtime(true);
 284
 285                 if (self::blockedURL($url)) {
 286                         logger('post_url: domain of ' . $url . ' is blocked', LOGGER_DATA);
 287                         return false;
 288                 }
 289
 290                 $a = get_app();
 291                 $ch = curl_init($url);
 292
 293                 if (($redirects > 8) || (!$ch)) {
 294                         return false;
 295                 }
 296
 297                 logger('post_url: start ' . $url, LOGGER_DATA);
 298
 299                 curl_setopt($ch, CURLOPT_HEADER, true);
 300                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 301                 curl_setopt($ch, CURLOPT_POST, 1);
 302                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 303                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 304
 305                 if (Config::get('system', 'ipv4_resolve', false)) {
 306                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 307                 }
 308
 309                 if (intval($timeout)) {
 310                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 311                 } else {
 312                         $curl_time = Config::get('system', 'curl_timeout', 60);
 313                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 314                 }
 315
 316                 if (defined('LIGHTTPD')) {
 317                         if (!is_array($headers)) {
 318                                 $headers = ['Expect:'];
 319                         } else {
 320                                 if (!in_array('Expect:', $headers)) {
 321                                         array_push($headers, 'Expect:');
 322                                 }
 323                         }
 324                 }
 325
 326                 if ($headers) {
 327                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 328                 }
 329
 330                 $check_cert = Config::get('system', 'verifyssl');
 331                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 332
 333                 if ($check_cert) {
 334                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 335                 }
 336
 337                 $proxy = Config::get('system', 'proxy');
 338
 339                 if (strlen($proxy)) {
 340                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 341                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 342                         $proxyuser = Config::get('system', 'proxyuser');
 343                         if (strlen($proxyuser)) {
 344                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 345                         }
 346                 }
 347
 348                 $a->set_curl_code(0);
 349
 350                 // don't let curl abort the entire application
 351                 // if it throws any errors.
 352
 353                 $s = @curl_exec($ch);
 354
 355                 $base = $s;
 356                 $curl_info = curl_getinfo($ch);
 357                 $http_code = $curl_info['http_code'];
 358
 359                 logger('post_url: result ' . $http_code . ' - ' . $url, LOGGER_DATA);
 360
 361                 $header = '';
 362
 363                 // Pull out multiple headers, e.g. proxy and continuation headers
 364                 // allow for HTTP/2.x without fixing code
 365
 366                 while (preg_match('/^HTTP\/[1-2].+? [1-5][0-9][0-9]/', $base)) {
 367                         $chunk = substr($base, 0, strpos($base, "\r\n\r\n") + 4);
 368                         $header .= $chunk;
 369                         $base = substr($base, strlen($chunk));
 370                 }
 371
 372                 if ($http_code == 301 || $http_code == 302 || $http_code == 303 || $http_code == 307) {
 373                         $matches = [];
 374                         preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
 375                         $newurl = trim(array_pop($matches));
 376
 377                         if (strpos($newurl, '/') === 0) {
 378                                 $newurl = $old_location_info["scheme"] . "://" . $old_location_info["host"] . $newurl;
 379                         }
 380
 381                         if (filter_var($newurl, FILTER_VALIDATE_URL)) {
 382                                 $redirects++;
 383                                 logger('post_url: redirect ' . $url . ' to ' . $newurl);
 384                                 return self::postURL($newurl, $params, $headers, $redirects, $timeout);
 385                         }
 386                 }
 387
 388                 $a->set_curl_code($http_code);
 389
 390                 $body = substr($s, strlen($header));
 391
 392                 $a->set_curl_headers($header);
 393
 394                 curl_close($ch);
 395
 396                 $a->save_timestamp($stamp1, 'network');
 397
 398                 logger('post_url: end ' . $url, LOGGER_DATA);
 399
 400                 return $body;
 401         }
 402
 403         /**
 404          * Generic XML return
 405          * Outputs a basic dfrn XML status structure to STDOUT, with a <status> variable
 406          * of $st and an optional text <message> of $message and terminates the current process.
 407          */
 408         public static function xmlStatus($st, $message = '')
 409         {
 410                 $result = ['status' => $st];
 411
 412                 if ($message != '') {
 413                         $result['message'] = $message;
 414                 }
 415
 416                 if ($st) {
 417                         logger('xml_status returning non_zero: ' . $st . " message=" . $message);
 418                 }
 419
 420                 header("Content-type: text/xml");
 421
 422                 $xmldata = ["result" => $result];
 423
 424                 echo XML::fromArray($xmldata, $xml);
 425
 426                 killme();
 427         }
 428
 429         /**
 430          * @brief Send HTTP status header and exit.
 431          *
 432          * @param integer $val         HTTP status result value
 433          * @param array   $description optional message
 434          *                             'title' => header title
 435          *                             'description' => optional message
 436          */
 437         public static function httpStatusExit($val, $description = [])
 438         {
 439                 $err = '';
 440                 if ($val >= 400) {
 441                         $err = 'Error';
 442                         if (!isset($description["title"])) {
 443                                 $description["title"] = $err." ".$val;
 444                         }
 445                 }
 446
 447                 if ($val >= 200 && $val < 300) {
 448                         $err = 'OK';
 449                 }
 450
 451                 logger('http_status_exit ' . $val);
 452                 header($_SERVER["SERVER_PROTOCOL"] . ' ' . $val . ' ' . $err);
 453
 454                 if (isset($description["title"])) {
 455                         $tpl = get_markup_template('http_status.tpl');
 456                         echo replace_macros(
 457                                 $tpl,
 458                                 [
 459                                         '$title' => $description["title"],
 460                                         '$description' => $description["description"]]
 461                         );
 462                 }
 463
 464                 killme();
 465         }
 466
 467         /**
 468          * @brief Check URL to se if ts's real
 469          *
 470          * Take a URL from the wild, prepend http:// if necessary
 471          * and check DNS to see if it's real (or check if is a valid IP address)
 472          *
 473          * @param string $url The URL to be validated
 474          * @return string|boolean The actual working URL, false else
 475          */
 476         public static function validateURL($url)
 477         {
 478                 if (Config::get('system', 'disable_url_validation')) {
 479                         return $url;
 480                 }
 481
 482                 // no naked subdomains (allow localhost for tests)
 483                 if (strpos($url, '.') === false && strpos($url, '/localhost/') === false) {
 484                         return false;
 485                 }
 486
 487                 if (substr($url, 0, 4) != 'http') {
 488                         $url = 'http://' . $url;
 489                 }
 490
 491                 /// @TODO Really suppress function outcomes? Why not find them + debug them?
 492                 $h = @parse_url($url);
 493
 494                 if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
 495                         return $url;
 496                 }
 497
 498                 return false;
 499         }
 500
 501         /**
 502          * @brief Checks that email is an actual resolvable internet address
 503          *
 504          * @param string $addr The email address
 505          * @return boolean True if it's a valid email address, false if it's not
 506          */
 507         public static function validateEmail($addr)
 508         {
 509                 if (Config::get('system', 'disable_email_validation')) {
 510                         return true;
 511                 }
 512
 513                 if (! strpos($addr, '@')) {
 514                         return false;
 515                 }
 516
 517                 $h = substr($addr, strpos($addr, '@') + 1);
 518
 519                 if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
 520                         return true;
 521                 }
 522                 return false;
 523         }
 524
 525         /**
 526          * @brief Check if URL is allowed
 527          *
 528          * Check $url against our list of allowed sites,
 529          * wildcards allowed. If allowed_sites is unset return true;
 530          *
 531          * @param string $url URL which get tested
 532          * @return boolean True if url is allowed otherwise return false
 533          */
 534         public static function allowedURL($url)
 535         {
 536                 $h = @parse_url($url);
 537
 538                 if (! $h) {
 539                         return false;
 540                 }
 541
 542                 $str_allowed = Config::get('system', 'allowed_sites');
 543                 if (! $str_allowed) {
 544                         return true;
 545                 }
 546
 547                 $found = false;
 548
 549                 $host = strtolower($h['host']);
 550
 551                 // always allow our own site
 552                 if ($host == strtolower($_SERVER['SERVER_NAME'])) {
 553                         return true;
 554                 }
 555
 556                 $fnmatch = function_exists('fnmatch');
 557                 $allowed = explode(',', $str_allowed);
 558
 559                 if (count($allowed)) {
 560                         foreach ($allowed as $a) {
 561                                 $pat = strtolower(trim($a));
 562                                 if (($fnmatch && fnmatch($pat, $host)) || ($pat == $host)) {
 563                                         $found = true;
 564                                         break;
 565                                 }
 566                         }
 567                 }
 568                 return $found;
 569         }
 570
 571         /**
 572          * Checks if the provided url domain is on the domain blocklist.
 573          * Returns true if it is or malformed URL, false if not.
 574          *
 575          * @param string $url The url to check the domain from
 576          *
 577          * @return boolean
 578          */
 579         public static function blockedURL($url)
 580         {
 581                 $h = @parse_url($url);
 582
 583                 if (! $h) {
 584                         return true;
 585                 }
 586
 587                 $domain_blocklist = Config::get('system', 'blocklist', []);
 588                 if (! $domain_blocklist) {
 589                         return false;
 590                 }
 591
 592                 $host = strtolower($h['host']);
 593
 594                 foreach ($domain_blocklist as $domain_block) {
 595                         if (strtolower($domain_block['domain']) == $host) {
 596                                 return true;
 597                         }
 598                 }
 599
 600                 return false;
 601         }
 602
 603         /**
 604          * @brief Check if email address is allowed to register here.
 605          *
 606          * Compare against our list (wildcards allowed).
 607          *
 608          * @param  string $email email address
 609          * @return boolean False if not allowed, true if allowed
 610          *    or if allowed list is not configured
 611          */
 612         public static function allowedEmail($email)
 613         {
 614                 $domain = strtolower(substr($email, strpos($email, '@') + 1));
 615                 if (!$domain) {
 616                         return false;
 617                 }
 618
 619                 $str_allowed = Config::get('system', 'allowed_email', '');
 620                 if (!x($str_allowed)) {
 621                         return true;
 622                 }
 623
 624                 $allowed = explode(',', $str_allowed);
 625
 626                 return self::allowedDomain($domain, $allowed);
 627         }
 628
 629         /**
 630          * Checks for the existence of a domain in a domain list
 631          *
 632          * @brief Checks for the existence of a domain in a domain list
 633          * @param string $domain
 634          * @param array  $domain_list
 635          * @return boolean
 636          */
 637         public static function allowedDomain($domain, array $domain_list)
 638         {
 639                 $found = false;
 640
 641                 foreach ($domain_list as $item) {
 642                         $pat = strtolower(trim($item));
 643                         if (fnmatch($pat, $domain) || ($pat == $domain)) {
 644                                 $found = true;
 645                                 break;
 646                         }
 647                 }
 648
 649                 return $found;
 650         }
 651
 652         public static function avatarImg($email)
 653         {
 654                 $avatar['size'] = 175;
 655                 $avatar['email'] = $email;
 656                 $avatar['url'] = '';
 657                 $avatar['success'] = false;
 658
 659                 Addon::callHooks('avatar_lookup', $avatar);
 660
 661                 if (! $avatar['success']) {
 662                         $avatar['url'] = System::baseUrl() . '/images/person-175.jpg';
 663                 }
 664
 665                 logger('Avatar: ' . $avatar['email'] . ' ' . $avatar['url'], LOGGER_DEBUG);
 666                 return $avatar['url'];
 667         }
 668
 669         public static function parseXmlString($s, $strict = true)
 670         {
 671                 // the "strict" parameter is deactivated
 672
 673                 /// @todo Move this function to the xml class
 674                 libxml_use_internal_errors(true);
 675
 676                 $x = @simplexml_load_string($s);
 677                 if (!$x) {
 678                         logger('libxml: parse: error: ' . $s, LOGGER_DATA);
 679                         foreach (libxml_get_errors() as $err) {
 680                                 logger('libxml: parse: ' . $err->code." at ".$err->line.":".$err->column." : ".$err->message, LOGGER_DATA);
 681                         }
 682                         libxml_clear_errors();
 683                 }
 684                 return $x;
 685         }
 686
 687         public static function scaleExternalImages($srctext, $include_link = true, $scale_replace = false)
 688         {
 689                 // Suppress "view full size"
 690                 if (intval(Config::get('system', 'no_view_full_size'))) {
 691                         $include_link = false;
 692                 }
 693
 694                 // Picture addresses can contain special characters
 695                 $s = htmlspecialchars_decode($srctext);
 696
 697                 $matches = null;
 698                 $c = preg_match_all('/\[img.*?\](.*?)\[\/img\]/ism', $s, $matches, PREG_SET_ORDER);
 699                 if ($c) {
 700                         foreach ($matches as $mtch) {
 701                                 logger('scale_external_image: ' . $mtch[1]);
 702
 703                                 $hostname = str_replace('www.', '', substr(System::baseUrl(), strpos(System::baseUrl(), '://') + 3));
 704                                 if (stristr($mtch[1], $hostname)) {
 705                                         continue;
 706                                 }
 707
 708                                 // $scale_replace, if passed, is an array of two elements. The
 709                                 // first is the name of the full-size image. The second is the
 710                                 // name of a remote, scaled-down version of the full size image.
 711                                 // This allows Friendica to display the smaller remote image if
 712                                 // one exists, while still linking to the full-size image
 713                                 if ($scale_replace) {
 714                                         $scaled = str_replace($scale_replace[0], $scale_replace[1], $mtch[1]);
 715                                 } else {
 716                                         $scaled = $mtch[1];
 717                                 }
 718                                 $i = self::fetchURL($scaled);
 719                                 if (! $i) {
 720                                         return $srctext;
 721                                 }
 722
 723                                 // guess mimetype from headers or filename
 724                                 $type = Image::guessType($mtch[1], true);
 725
 726                                 if ($i) {
 727                                         $Image = new Image($i, $type);
 728                                         if ($Image->isValid()) {
 729                                                 $orig_width = $Image->getWidth();
 730                                                 $orig_height = $Image->getHeight();
 731
 732                                                 if ($orig_width > 640 || $orig_height > 640) {
 733                                                         $Image->scaleDown(640);
 734                                                         $new_width = $Image->getWidth();
 735                                                         $new_height = $Image->getHeight();
 736                                                         logger('scale_external_images: ' . $orig_width . '->' . $new_width . 'w ' . $orig_height . '->' . $new_height . 'h' . ' match: ' . $mtch[0], LOGGER_DEBUG);
 737                                                         $s = str_replace(
 738                                                                 $mtch[0],
 739                                                                 '[img=' . $new_width . 'x' . $new_height. ']' . $scaled . '[/img]'
 740                                                                 . "\n" . (($include_link)
 741                                                                         ? '[url=' . $mtch[1] . ']' . L10n::t('view full size') . '[/url]' . "\n"
 742                                                                         : ''),
 743                                                                 $s
 744                                                         );
 745                                                         logger('scale_external_images: new string: ' . $s, LOGGER_DEBUG);
 746                                                 }
 747                                         }
 748                                 }
 749                         }
 750                 }
 751
 752                 // replace the special char encoding
 753                 $s = htmlspecialchars($s, ENT_NOQUOTES, 'UTF-8');
 754                 return $s;
 755         }
 756
 757         public static function fixContactSslPolicy(&$contact, $new_policy)
 758         {
 759                 $ssl_changed = false;
 760                 if ((intval($new_policy) == SSL_POLICY_SELFSIGN || $new_policy === 'self') && strstr($contact['url'], 'https:')) {
 761                         $ssl_changed = true;
 762                         $contact['url']     =   str_replace('https:', 'http:', $contact['url']);
 763                         $contact['request'] =   str_replace('https:', 'http:', $contact['request']);
 764                         $contact['notify']  =   str_replace('https:', 'http:', $contact['notify']);
 765                         $contact['poll']    =   str_replace('https:', 'http:', $contact['poll']);
 766                         $contact['confirm'] =   str_replace('https:', 'http:', $contact['confirm']);
 767                         $contact['poco']    =   str_replace('https:', 'http:', $contact['poco']);
 768                 }
 769
 770                 if ((intval($new_policy) == SSL_POLICY_FULL || $new_policy === 'full') && strstr($contact['url'], 'http:')) {
 771                         $ssl_changed = true;
 772                         $contact['url']     =   str_replace('http:', 'https:', $contact['url']);
 773                         $contact['request'] =   str_replace('http:', 'https:', $contact['request']);
 774                         $contact['notify']  =   str_replace('http:', 'https:', $contact['notify']);
 775                         $contact['poll']    =   str_replace('http:', 'https:', $contact['poll']);
 776                         $contact['confirm'] =   str_replace('http:', 'https:', $contact['confirm']);
 777                         $contact['poco']    =   str_replace('http:', 'https:', $contact['poco']);
 778                 }
 779
 780                 if ($ssl_changed) {
 781                         $fields = ['url' => $contact['url'], 'request' => $contact['request'],
 782                                         'notify' => $contact['notify'], 'poll' => $contact['poll'],
 783                                         'confirm' => $contact['confirm'], 'poco' => $contact['poco']];
 784                         dba::update('contact', $fields, ['id' => $contact['id']]);
 785                 }
 786         }
 787
 788         /**
 789          * @brief Remove Google Analytics and other tracking platforms params from URL
 790          *
 791          * @param string $url Any user-submitted URL that may contain tracking params
 792          * @return string The same URL stripped of tracking parameters
 793          */
 794         public static function stripTrackingQueryParams($url)
 795         {
 796                 $urldata = parse_url($url);
 797                 if (is_string($urldata["query"])) {
 798                         $query = $urldata["query"];
 799                         parse_str($query, $querydata);
 800
 801                         if (is_array($querydata)) {
 802                                 foreach ($querydata as $param => $value) {
 803                                         if (in_array(
 804                                                 $param,
 805                                                 [
 806                                                         "utm_source", "utm_medium", "utm_term", "utm_content", "utm_campaign",
 807                                                         "wt_mc", "pk_campaign", "pk_kwd", "mc_cid", "mc_eid",
 808                                                         "fb_action_ids", "fb_action_types", "fb_ref",
 809                                                         "awesm", "wtrid",
 810                                                         "woo_campaign", "woo_source", "woo_medium", "woo_content", "woo_term"]
 811                                                 )
 812                                         ) {
 813                                                 $pair = $param . "=" . urlencode($value);
 814                                                 $url = str_replace($pair, "", $url);
 815
 816                                                 // Second try: if the url isn't encoded completely
 817                                                 $pair = $param . "=" . str_replace(" ", "+", $value);
 818                                                 $url = str_replace($pair, "", $url);
 819
 820                                                 // Third try: Maybey the url isn't encoded at all
 821                                                 $pair = $param . "=" . $value;
 822                                                 $url = str_replace($pair, "", $url);
 823
 824                                                 $url = str_replace(["?&", "&&"], ["?", ""], $url);
 825                                         }
 826                                 }
 827                         }
 828
 829                         if (substr($url, -1, 1) == "?") {
 830                                 $url = substr($url, 0, -1);
 831                         }
 832                 }
 833
 834                 return $url;
 835         }
 836
 837         /**
 838          * @brief Returns the original URL of the provided URL
 839          *
 840          * This function strips tracking query params and follows redirections, either
 841          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 842          *
 843          * @todo Remove the $fetchbody parameter that generates an extraneous HEAD request
 844          *
 845          * @see ParseUrl::getSiteinfo
 846          *
 847          * @param string $url       A user-submitted URL
 848          * @param int    $depth     The current redirection recursion level (internal)
 849          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 850          * @return string A canonical URL
 851          */
 852         public static function originalURL($url, $depth = 1, $fetchbody = false)
 853         {
 854                 $a = get_app();
 855
 856                 $url = self::stripTrackingQueryParams($url);
 857
 858                 if ($depth > 10) {
 859                         return($url);
 860                 }
 861
 862                 $url = trim($url, "'");
 863
 864                 $stamp1 = microtime(true);
 865
 866                 $ch = curl_init();
 867                 curl_setopt($ch, CURLOPT_URL, $url);
 868                 curl_setopt($ch, CURLOPT_HEADER, 1);
 869                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 870                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 871                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 872                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 873
 874                 curl_exec($ch);
 875                 $curl_info = @curl_getinfo($ch);
 876                 $http_code = $curl_info['http_code'];
 877                 curl_close($ch);
 878
 879                 $a->save_timestamp($stamp1, "network");
 880
 881                 if ($http_code == 0) {
 882                         return($url);
 883                 }
 884
 885                 if ((($curl_info['http_code'] == "301") || ($curl_info['http_code'] == "302"))
 886                         && (($curl_info['redirect_url'] != "") || ($curl_info['location'] != ""))
 887                 ) {
 888                         if ($curl_info['redirect_url'] != "") {
 889                                 return(Network::originalURL($curl_info['redirect_url'], ++$depth, $fetchbody));
 890                         } else {
 891                                 return(Network::originalURL($curl_info['location'], ++$depth, $fetchbody));
 892                         }
 893                 }
 894
 895                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 896                 if (!$fetchbody) {
 897                         return(Network::originalURL($url, ++$depth, true));
 898                 }
 899
 900                 // if the file is too large then exit
 901                 if ($curl_info["download_content_length"] > 1000000) {
 902                         return($url);
 903                 }
 904
 905                 // if it isn't a HTML file then exit
 906                 if (($curl_info["content_type"] != "") && !strstr(strtolower($curl_info["content_type"]), "html")) {
 907                         return($url);
 908                 }
 909
 910                 $stamp1 = microtime(true);
 911
 912                 $ch = curl_init();
 913                 curl_setopt($ch, CURLOPT_URL, $url);
 914                 curl_setopt($ch, CURLOPT_HEADER, 0);
 915                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 916                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 917                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 918                 curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
 919
 920                 $body = curl_exec($ch);
 921                 curl_close($ch);
 922
 923                 $a->save_timestamp($stamp1, "network");
 924
 925                 if (trim($body) == "") {
 926                         return($url);
 927                 }
 928
 929                 // Check for redirect in meta elements
 930                 $doc = new DOMDocument();
 931                 @$doc->loadHTML($body);
 932
 933                 $xpath = new DomXPath($doc);
 934
 935                 $list = $xpath->query("//meta[@content]");
 936                 foreach ($list as $node) {
 937                         $attr = [];
 938                         if ($node->attributes->length) {
 939                                 foreach ($node->attributes as $attribute) {
 940                                         $attr[$attribute->name] = $attribute->value;
 941                                 }
 942                         }
 943
 944                         if (@$attr["http-equiv"] == 'refresh') {
 945                                 $path = $attr["content"];
 946                                 $pathinfo = explode(";", $path);
 947                                 foreach ($pathinfo as $value) {
 948                                         if (substr(strtolower($value), 0, 4) == "url=") {
 949                                                 return(Network::originalURL(substr($value, 4), ++$depth));
 950                                         }
 951                                 }
 952                         }
 953                 }
 954
 955                 return $url;
 956         }
 957
 958         public static function shortLink($url)
 959         {
 960                 $slinky = new Slinky($url);
 961                 $yourls_url = Config::get('yourls', 'url1');
 962                 if ($yourls_url) {
 963                         $yourls_username = Config::get('yourls', 'username1');
 964                         $yourls_password = Config::get('yourls', 'password1');
 965                         $yourls_ssl = Config::get('yourls', 'ssl1');
 966                         $yourls = new Slinky_YourLS();
 967                         $yourls->set('username', $yourls_username);
 968                         $yourls->set('password', $yourls_password);
 969                         $yourls->set('ssl', $yourls_ssl);
 970                         $yourls->set('yourls-url', $yourls_url);
 971                         $slinky->set_cascade([$yourls, new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 972                 } else {
 973                         // setup a cascade of shortening services
 974                         // try to get a short link from these services
 975                         // in the order ur1.ca, tinyurl
 976                         $slinky->set_cascade([new Slinky_Ur1ca(), new Slinky_TinyURL()]);
 977                 }
 978                 return $slinky->short();
 979         }
 980
 981         /**
 982          * @brief Encodes content to json
 983          *
 984          * This function encodes an array to json format
 985          * and adds an application/json HTTP header to the output.
 986          * After finishing the process is getting killed.
 987          *
 988          * @param array $x The input content
 989          */
 990         public static function jsonReturnAndDie($x)
 991         {
 992                 header("content-type: application/json");
 993                 echo json_encode($x);
 994                 killme();
 995         }
 996
 997         /**
 998          * @brief Find the matching part between two url
 999          *
1000          * @param string $url1
1001          * @param string $url2
1002          * @return string The matching part
1003          */
1004         public static function matchingURL($url1, $url2)
1005         {
1006                 if (($url1 == "") || ($url2 == "")) {
1007                         return "";
1008                 }
1009
1010                 $url1 = normalise_link($url1);
1011                 $url2 = normalise_link($url2);
1012
1013                 $parts1 = parse_url($url1);
1014                 $parts2 = parse_url($url2);
1015
1016                 if (!isset($parts1["host"]) || !isset($parts2["host"])) {
1017                         return "";
1018                 }
1019
1020                 if ($parts1["scheme"] != $parts2["scheme"]) {
1021                         return "";
1022                 }
1023
1024                 if ($parts1["host"] != $parts2["host"]) {
1025                         return "";
1026                 }
1027
1028                 if ($parts1["port"] != $parts2["port"]) {
1029                         return "";
1030                 }
1031
1032                 $match = $parts1["scheme"]."://".$parts1["host"];
1033
1034                 if ($parts1["port"]) {
1035                         $match .= ":".$parts1["port"];
1036                 }
1037
1038                 $pathparts1 = explode("/", $parts1["path"]);
1039                 $pathparts2 = explode("/", $parts2["path"]);
1040
1041                 $i = 0;
1042                 $path = "";
1043                 do {
1044                         $path1 = $pathparts1[$i];
1045                         $path2 = $pathparts2[$i];
1046
1047                         if ($path1 == $path2) {
1048                                 $path .= $path1."/";
1049                         }
1050                 } while (($path1 == $path2) && ($i++ <= count($pathparts1)));
1051
1052                 $match .= $path;
1053
1054                 return normalise_link($match);
1055         }
1056
1057         /**
1058          * @brief Glue url parts together
1059          *
1060          * @param array $parsed URL parts
1061          *
1062          * @return string The glued URL
1063          */
1064         public static function unParseURL($parsed)
1065         {
1066                 $get = function ($key) use ($parsed) {
1067                         return isset($parsed[$key]) ? $parsed[$key] : null;
1068                 };
1069
1070                 $pass      = $get('pass');
1071                 $user      = $get('user');
1072                 $userinfo  = $pass !== null ? "$user:$pass" : $user;
1073                 $port      = $get('port');
1074                 $scheme    = $get('scheme');
1075                 $query     = $get('query');
1076                 $fragment  = $get('fragment');
1077                 $authority = ($userinfo !== null ? $userinfo."@" : '') .
1078                                                 $get('host') .
1079                                                 ($port ? ":$port" : '');
1080
1081                 return  (strlen($scheme) ? $scheme.":" : '') .
1082                         (strlen($authority) ? "//".$authority : '') .
1083                         $get('path') .
1084                         (strlen($query) ? "?".$query : '') .
1085                         (strlen($fragment) ? "#".$fragment : '');
1086         }
1087 }