includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 use MediaWiki\MediaWikiServices;
  27 use MediaWiki\Session\Session;
  28 use MediaWiki\Session\SessionId;
  29 use MediaWiki\Session\SessionManager;
  30 use Wikimedia\AtEase\AtEase;
  31
  32 // The point of this class is to be a wrapper around super globals
  33 // phpcs:disable MediaWiki.Usage.SuperGlobalsUsage.SuperGlobals
  34
  35 /**
  36  * The WebRequest class encapsulates getting at data passed in the
  37  * URL or via a POSTed form stripping illegal input characters and
  38  * normalizing Unicode sequences.
  39  *
  40  * @ingroup HTTP
  41  */
  42 class WebRequest {
  43         /** @var array */
  44         protected $data;
  45         /** @var array */
  46         protected $headers = [];
  47
  48         /**
  49          * Flag to make WebRequest::getHeader return an array of values.
  50          * @since 1.26
  51          */
  52         const GETHEADER_LIST = 1;
  53
  54         /**
  55          * The unique request ID.
  56          * @var string
  57          */
  58         private static $reqId;
  59
  60         /**
  61          * Lazy-init response object
  62          * @var WebResponse
  63          */
  64         private $response;
  65
  66         /**
  67          * Cached client IP address
  68          * @var string
  69          */
  70         private $ip;
  71
  72         /**
  73          * The timestamp of the start of the request, with microsecond precision.
  74          * @var float
  75          */
  76         protected $requestTime;
  77
  78         /**
  79          * Cached URL protocol
  80          * @var string
  81          */
  82         protected $protocol;
  83
  84         /**
  85          * @var SessionId|null Session ID to use for this
  86          *  request. We can't save the session directly due to reference cycles not
  87          *  working too well (slow GC in Zend and never collected in HHVM).
  88          */
  89         protected $sessionId = null;
  90
  91         /** @var bool Whether this HTTP request is "safe" (even if it is an HTTP post) */
  92         protected $markedAsSafe = false;
  93
  94         /**
  95          * @codeCoverageIgnore
  96          */
  97         public function __construct() {
  98                 $this->requestTime = $_SERVER['REQUEST_TIME_FLOAT'];
  99
 100                 // POST overrides GET data
 101                 // We don't use $_REQUEST here to avoid interference from cookies...
 102                 $this->data = $_POST + $_GET;
 103         }
 104
 105         /**
 106          * Extract relevant query arguments from the http request uri's path
 107          * to be merged with the normal php provided query arguments.
 108          * Tries to use the REQUEST_URI data if available and parses it
 109          * according to the wiki's configuration looking for any known pattern.
 110          *
 111          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
 112          * provided by the server if any and use that to set a 'title' parameter.
 113          *
 114          * @param string $want If this is not 'all', then the function
 115          * will return an empty array if it determines that the URL is
 116          * inside a rewrite path.
 117          *
 118          * @return array Any query arguments found in path matches.
 119          */
 120         public static function getPathInfo( $want = 'all' ) {
 121                 // PATH_INFO is mangled due to https://bugs.php.net/bug.php?id=31892
 122                 // And also by Apache 2.x, double slashes are converted to single slashes.
 123                 // So we will use REQUEST_URI if possible.
 124                 if ( isset( $_SERVER['REQUEST_URI'] ) ) {
 125                         // Slurp out the path portion to examine...
 126                         $url = $_SERVER['REQUEST_URI'];
 127                         if ( !preg_match( '!^https?://!', $url ) ) {
 128                                 $url = 'http://unused' . $url;
 129                         }
 130                         AtEase::suppressWarnings();
 131                         $a = parse_url( $url );
 132                         AtEase::restoreWarnings();
 133                         if ( !$a ) {
 134                                 return [];
 135                         }
 136                         $path = $a['path'] ?? '';
 137
 138                         global $wgScript;
 139                         if ( $path == $wgScript && $want !== 'all' ) {
 140                                 // Script inside a rewrite path?
 141                                 // Abort to keep from breaking...
 142                                 return [];
 143                         }
 144
 145                         $router = new PathRouter;
 146
 147                         // Raw PATH_INFO style
 148                         $router->add( "$wgScript/$1" );
 149
 150                         if ( isset( $_SERVER['SCRIPT_NAME'] )
 151                                 && strpos( $_SERVER['SCRIPT_NAME'], '.php' ) !== false
 152                         ) {
 153                                 // Check for SCRIPT_NAME, we handle index.php explicitly
 154                                 // But we do have some other .php files such as img_auth.php
 155                                 // Don't let root article paths clober the parsing for them
 156                                 $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 157                         }
 158
 159                         global $wgArticlePath;
 160                         if ( $wgArticlePath ) {
 161                                 $router->add( $wgArticlePath );
 162                         }
 163
 164                         global $wgActionPaths;
 165                         $articlePaths = PathRouter::getActionPaths( $wgActionPaths, $wgArticlePath );
 166                         if ( $articlePaths ) {
 167                                 $router->add( $articlePaths, [ 'action' => '$key' ] );
 168                         }
 169
 170                         global $wgVariantArticlePath;
 171                         if ( $wgVariantArticlePath ) {
 172                                 $router->add( $wgVariantArticlePath,
 173                                         [ 'variant' => '$2' ],
 174                                         [ '$2' => MediaWikiServices::getInstance()->getContentLanguage()->
 175                                         getVariants() ]
 176                                 );
 177                         }
 178
 179                         Hooks::run( 'WebRequestPathInfoRouter', [ $router ] );
 180
 181                         $matches = $router->parse( $path );
 182                 } else {
 183                         global $wgUsePathInfo;
 184                         $matches = [];
 185                         if ( $wgUsePathInfo ) {
 186                                 if ( !empty( $_SERVER['ORIG_PATH_INFO'] ) ) {
 187                                         // Mangled PATH_INFO
 188                                         // https://bugs.php.net/bug.php?id=31892
 189                                         // Also reported when ini_get('cgi.fix_pathinfo')==false
 190                                         $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 191                                 } elseif ( !empty( $_SERVER['PATH_INFO'] ) ) {
 192                                         // Regular old PATH_INFO yay
 193                                         $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 194                                 }
 195                         }
 196                 }
 197
 198                 return $matches;
 199         }
 200
 201         /**
 202          * Work out an appropriate URL prefix containing scheme and host, based on
 203          * information detected from $_SERVER
 204          *
 205          * @return string
 206          */
 207         public static function detectServer() {
 208                 global $wgAssumeProxiesUseDefaultProtocolPorts;
 209
 210                 $proto = self::detectProtocol();
 211                 $stdPort = $proto === 'https' ? 443 : 80;
 212
 213                 $varNames = [ 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' ];
 214                 $host = 'localhost';
 215                 $port = $stdPort;
 216                 foreach ( $varNames as $varName ) {
 217                         if ( !isset( $_SERVER[$varName] ) ) {
 218                                 continue;
 219                         }
 220
 221                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 222                         if ( !$parts ) {
 223                                 // Invalid, do not use
 224                                 continue;
 225                         }
 226
 227                         $host = $parts[0];
 228                         if ( $wgAssumeProxiesUseDefaultProtocolPorts && isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) ) {
 229                                 // T72021: Assume that upstream proxy is running on the default
 230                                 // port based on the protocol. We have no reliable way to determine
 231                                 // the actual port in use upstream.
 232                                 $port = $stdPort;
 233                         } elseif ( $parts[1] === false ) {
 234                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 235                                         $port = $_SERVER['SERVER_PORT'];
 236                                 } // else leave it as $stdPort
 237                         } else {
 238                                 $port = $parts[1];
 239                         }
 240                         break;
 241                 }
 242
 243                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 244         }
 245
 246         /**
 247          * Detect the protocol from $_SERVER.
 248          * This is for use prior to Setup.php, when no WebRequest object is available.
 249          * At other times, use the non-static function getProtocol().
 250          *
 251          * @return string
 252          */
 253         public static function detectProtocol() {
 254                 if ( ( !empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) ||
 255                         ( isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) &&
 256                         $_SERVER['HTTP_X_FORWARDED_PROTO'] === 'https' ) ) {
 257                         return 'https';
 258                 } else {
 259                         return 'http';
 260                 }
 261         }
 262
 263         /**
 264          * Get the number of seconds to have elapsed since request start,
 265          * in fractional seconds, with microsecond resolution.
 266          *
 267          * @return float
 268          * @since 1.25
 269          */
 270         public function getElapsedTime() {
 271                 return microtime( true ) - $this->requestTime;
 272         }
 273
 274         /**
 275          * Get the unique request ID.
 276          * This is either the value of the UNIQUE_ID envvar (if present) or a
 277          * randomly-generated 24-character string.
 278          *
 279          * @return string
 280          * @since 1.27
 281          */
 282         public static function getRequestId() {
 283                 // This method is called from various error handlers and should be kept simple.
 284
 285                 if ( self::$reqId ) {
 286                         return self::$reqId;
 287                 }
 288
 289                 global $wgAllowExternalReqID;
 290
 291                 self::$reqId = $_SERVER['UNIQUE_ID'] ?? wfRandomString( 24 );
 292                 if ( $wgAllowExternalReqID ) {
 293                         $id = RequestContext::getMain()->getRequest()->getHeader( 'X-Request-Id' );
 294                         if ( $id ) {
 295                                 self::$reqId = $id;
 296                         }
 297                 }
 298
 299                 return self::$reqId;
 300         }
 301
 302         /**
 303          * Override the unique request ID. This is for sub-requests, such as jobs,
 304          * that wish to use the same id but are not part of the same execution context.
 305          *
 306          * @param string $id
 307          * @since 1.27
 308          */
 309         public static function overrideRequestId( $id ) {
 310                 self::$reqId = $id;
 311         }
 312
 313         /**
 314          * Get the current URL protocol (http or https)
 315          * @return string
 316          */
 317         public function getProtocol() {
 318                 if ( $this->protocol === null ) {
 319                         $this->protocol = self::detectProtocol();
 320                 }
 321                 return $this->protocol;
 322         }
 323
 324         /**
 325          * Check for title, action, and/or variant data in the URL
 326          * and interpolate it into the GET variables.
 327          * This should only be run after the content language is available,
 328          * as we may need the list of language variants to determine
 329          * available variant URLs.
 330          */
 331         public function interpolateTitle() {
 332                 // T18019: title interpolation on API queries is useless and sometimes harmful
 333                 if ( defined( 'MW_API' ) ) {
 334                         return;
 335                 }
 336
 337                 $matches = self::getPathInfo( 'title' );
 338                 foreach ( $matches as $key => $val ) {
 339                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 340                 }
 341         }
 342
 343         /**
 344          * URL rewriting function; tries to extract page title and,
 345          * optionally, one other fixed parameter value from a URL path.
 346          *
 347          * @param string $path The URL path given from the client
 348          * @param array $bases One or more URLs, optionally with $1 at the end
 349          * @param string|bool $key If provided, the matching key in $bases will be
 350          *    passed on as the value of this URL parameter
 351          * @return array Array of URL variables to interpolate; empty if no match
 352          */
 353         static function extractTitle( $path, $bases, $key = false ) {
 354                 foreach ( (array)$bases as $keyValue => $base ) {
 355                         // Find the part after $wgArticlePath
 356                         $base = str_replace( '$1', '', $base );
 357                         $baseLen = strlen( $base );
 358                         if ( substr( $path, 0, $baseLen ) == $base ) {
 359                                 $raw = substr( $path, $baseLen );
 360                                 if ( $raw !== '' ) {
 361                                         $matches = [ 'title' => rawurldecode( $raw ) ];
 362                                         if ( $key ) {
 363                                                 $matches[$key] = $keyValue;
 364                                         }
 365                                         return $matches;
 366                                 }
 367                         }
 368                 }
 369                 return [];
 370         }
 371
 372         /**
 373          * Recursively normalizes UTF-8 strings in the given array.
 374          *
 375          * @param string|array $data
 376          * @return array|string Cleaned-up version of the given
 377          * @private
 378          */
 379         public function normalizeUnicode( $data ) {
 380                 if ( is_array( $data ) ) {
 381                         foreach ( $data as $key => $val ) {
 382                                 $data[$key] = $this->normalizeUnicode( $val );
 383                         }
 384                 } else {
 385                         $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 386                         $data = $contLang ? $contLang->normalize( $data ) :
 387                                 UtfNormal\Validator::cleanUp( $data );
 388                 }
 389                 return $data;
 390         }
 391
 392         /**
 393          * Fetch a value from the given array or return $default if it's not set.
 394          *
 395          * @param array $arr
 396          * @param string $name
 397          * @param mixed $default
 398          * @return mixed
 399          */
 400         private function getGPCVal( $arr, $name, $default ) {
 401                 # PHP is so nice to not touch input data, except sometimes:
 402                 # https://www.php.net/variables.external#language.variables.external.dot-in-names
 403                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 404                 $name = strtr( $name, '.', '_' );
 405
 406                 if ( !isset( $arr[$name] ) ) {
 407                         return $default;
 408                 }
 409
 410                 $data = $arr[$name];
 411                 # Optimisation: Skip UTF-8 normalization and legacy transcoding for simple ASCII strings.
 412                 $isAsciiStr = ( is_string( $data ) && preg_match( '/[^\x20-\x7E]/', $data ) === 0 );
 413                 if ( !$isAsciiStr ) {
 414                         if ( isset( $_GET[$name] ) && is_string( $data ) ) {
 415                                 # Check for alternate/legacy character encoding.
 416                                 $data = MediaWikiServices::getInstance()
 417                                         ->getContentLanguage()
 418                                         ->checkTitleEncoding( $data );
 419                         }
 420                         $data = $this->normalizeUnicode( $data );
 421                 }
 422
 423                 return $data;
 424         }
 425
 426         /**
 427          * Fetch a scalar from the input without normalization, or return $default
 428          * if it's not set.
 429          *
 430          * Unlike self::getVal(), this does not perform any normalization on the
 431          * input value.
 432          *
 433          * @since 1.28
 434          * @param string $name
 435          * @param string|null $default
 436          * @return string|null
 437          */
 438         public function getRawVal( $name, $default = null ) {
 439                 $name = strtr( $name, '.', '_' ); // See comment in self::getGPCVal()
 440                 if ( isset( $this->data[$name] ) && !is_array( $this->data[$name] ) ) {
 441                         $val = $this->data[$name];
 442                 } else {
 443                         $val = $default;
 444                 }
 445                 if ( is_null( $val ) ) {
 446                         return $val;
 447                 } else {
 448                         return (string)$val;
 449                 }
 450         }
 451
 452         /**
 453          * Fetch a scalar from the input or return $default if it's not set.
 454          * Returns a string. Arrays are discarded. Useful for
 455          * non-freeform text inputs (e.g. predefined internal text keys
 456          * selected by a drop-down menu). For freeform input, see getText().
 457          *
 458          * @param string $name
 459          * @param string|null $default Optional default (or null)
 460          * @return string|null
 461          */
 462         public function getVal( $name, $default = null ) {
 463                 $val = $this->getGPCVal( $this->data, $name, $default );
 464                 if ( is_array( $val ) ) {
 465                         $val = $default;
 466                 }
 467                 if ( is_null( $val ) ) {
 468                         return $val;
 469                 } else {
 470                         return (string)$val;
 471                 }
 472         }
 473
 474         /**
 475          * Set an arbitrary value into our get/post data.
 476          *
 477          * @param string $key Key name to use
 478          * @param mixed $value Value to set
 479          * @return mixed Old value if one was present, null otherwise
 480          */
 481         public function setVal( $key, $value ) {
 482                 $ret = $this->data[$key] ?? null;
 483                 $this->data[$key] = $value;
 484                 return $ret;
 485         }
 486
 487         /**
 488          * Unset an arbitrary value from our get/post data.
 489          *
 490          * @param string $key Key name to use
 491          * @return mixed Old value if one was present, null otherwise
 492          */
 493         public function unsetVal( $key ) {
 494                 if ( !isset( $this->data[$key] ) ) {
 495                         $ret = null;
 496                 } else {
 497                         $ret = $this->data[$key];
 498                         unset( $this->data[$key] );
 499                 }
 500                 return $ret;
 501         }
 502
 503         /**
 504          * Fetch an array from the input or return $default if it's not set.
 505          * If source was scalar, will return an array with a single element.
 506          * If no source and no default, returns null.
 507          *
 508          * @param string $name
 509          * @param array|null $default Optional default (or null)
 510          * @return array|null
 511          */
 512         public function getArray( $name, $default = null ) {
 513                 $val = $this->getGPCVal( $this->data, $name, $default );
 514                 if ( is_null( $val ) ) {
 515                         return null;
 516                 } else {
 517                         return (array)$val;
 518                 }
 519         }
 520
 521         /**
 522          * Fetch an array of integers, or return $default if it's not set.
 523          * If source was scalar, will return an array with a single element.
 524          * If no source and no default, returns null.
 525          * If an array is returned, contents are guaranteed to be integers.
 526          *
 527          * @param string $name
 528          * @param array|null $default Option default (or null)
 529          * @return array Array of ints
 530          */
 531         public function getIntArray( $name, $default = null ) {
 532                 $val = $this->getArray( $name, $default );
 533                 if ( is_array( $val ) ) {
 534                         $val = array_map( 'intval', $val );
 535                 }
 536                 return $val;
 537         }
 538
 539         /**
 540          * Fetch an integer value from the input or return $default if not set.
 541          * Guaranteed to return an integer; non-numeric input will typically
 542          * return 0.
 543          *
 544          * @param string $name
 545          * @param int $default
 546          * @return int
 547          */
 548         public function getInt( $name, $default = 0 ) {
 549                 return intval( $this->getRawVal( $name, $default ) );
 550         }
 551
 552         /**
 553          * Fetch an integer value from the input or return null if empty.
 554          * Guaranteed to return an integer or null; non-numeric input will
 555          * typically return null.
 556          *
 557          * @param string $name
 558          * @return int|null
 559          */
 560         public function getIntOrNull( $name ) {
 561                 $val = $this->getRawVal( $name );
 562                 return is_numeric( $val )
 563                         ? intval( $val )
 564                         : null;
 565         }
 566
 567         /**
 568          * Fetch a floating point value from the input or return $default if not set.
 569          * Guaranteed to return a float; non-numeric input will typically
 570          * return 0.
 571          *
 572          * @since 1.23
 573          * @param string $name
 574          * @param float $default
 575          * @return float
 576          */
 577         public function getFloat( $name, $default = 0.0 ) {
 578                 return floatval( $this->getRawVal( $name, $default ) );
 579         }
 580
 581         /**
 582          * Fetch a boolean value from the input or return $default if not set.
 583          * Guaranteed to return true or false, with normal PHP semantics for
 584          * boolean interpretation of strings.
 585          *
 586          * @param string $name
 587          * @param bool $default
 588          * @return bool
 589          */
 590         public function getBool( $name, $default = false ) {
 591                 return (bool)$this->getRawVal( $name, $default );
 592         }
 593
 594         /**
 595          * Fetch a boolean value from the input or return $default if not set.
 596          * Unlike getBool, the string "false" will result in boolean false, which is
 597          * useful when interpreting information sent from JavaScript.
 598          *
 599          * @param string $name
 600          * @param bool $default
 601          * @return bool
 602          */
 603         public function getFuzzyBool( $name, $default = false ) {
 604                 return $this->getBool( $name, $default )
 605                         && strcasecmp( $this->getRawVal( $name ), 'false' ) !== 0;
 606         }
 607
 608         /**
 609          * Return true if the named value is set in the input, whatever that
 610          * value is (even "0"). Return false if the named value is not set.
 611          * Example use is checking for the presence of check boxes in forms.
 612          *
 613          * @param string $name
 614          * @return bool
 615          */
 616         public function getCheck( $name ) {
 617                 # Checkboxes and buttons are only present when clicked
 618                 # Presence connotes truth, absence false
 619                 return $this->getRawVal( $name, null ) !== null;
 620         }
 621
 622         /**
 623          * Fetch a text string from the given array or return $default if it's not
 624          * set. Carriage returns are stripped from the text. This should generally
 625          * be used for form "<textarea>" and "<input>" fields, and for
 626          * user-supplied freeform text input.
 627          *
 628          * @param string $name
 629          * @param string $default Optional
 630          * @return string
 631          */
 632         public function getText( $name, $default = '' ) {
 633                 $val = $this->getVal( $name, $default );
 634                 return str_replace( "\r\n", "\n", $val );
 635         }
 636
 637         /**
 638          * Extracts the given named values into an array.
 639          * If no arguments are given, returns all input values.
 640          * No transformation is performed on the values.
 641          *
 642          * @return array
 643          */
 644         public function getValues() {
 645                 $names = func_get_args();
 646                 if ( count( $names ) == 0 ) {
 647                         $names = array_keys( $this->data );
 648                 }
 649
 650                 $retVal = [];
 651                 foreach ( $names as $name ) {
 652                         $value = $this->getGPCVal( $this->data, $name, null );
 653                         if ( !is_null( $value ) ) {
 654                                 $retVal[$name] = $value;
 655                         }
 656                 }
 657                 return $retVal;
 658         }
 659
 660         /**
 661          * Returns the names of all input values excluding those in $exclude.
 662          *
 663          * @param array $exclude
 664          * @return array
 665          */
 666         public function getValueNames( $exclude = [] ) {
 667                 return array_diff( array_keys( $this->getValues() ), $exclude );
 668         }
 669
 670         /**
 671          * Get the values passed in the query string.
 672          * No transformation is performed on the values.
 673          *
 674          * @codeCoverageIgnore
 675          * @return array
 676          */
 677         public function getQueryValues() {
 678                 return $_GET;
 679         }
 680
 681         /**
 682          * Get the values passed via POST.
 683          * No transformation is performed on the values.
 684          *
 685          * @since 1.32
 686          * @codeCoverageIgnore
 687          * @return array
 688          */
 689         public function getPostValues() {
 690                 return $_POST;
 691         }
 692
 693         /**
 694          * Return the contents of the Query with no decoding. Use when you need to
 695          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 696          *
 697          * @codeCoverageIgnore
 698          * @return string
 699          */
 700         public function getRawQueryString() {
 701                 return $_SERVER['QUERY_STRING'];
 702         }
 703
 704         /**
 705          * Return the contents of the POST with no decoding. Use when you need to
 706          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 707          *
 708          * @return string
 709          */
 710         public function getRawPostString() {
 711                 if ( !$this->wasPosted() ) {
 712                         return '';
 713                 }
 714                 return $this->getRawInput();
 715         }
 716
 717         /**
 718          * Return the raw request body, with no processing. Cached since some methods
 719          * disallow reading the stream more than once. As stated in the php docs, this
 720          * does not work with enctype="multipart/form-data".
 721          *
 722          * @return string
 723          */
 724         public function getRawInput() {
 725                 static $input = null;
 726                 if ( $input === null ) {
 727                         $input = file_get_contents( 'php://input' );
 728                 }
 729                 return $input;
 730         }
 731
 732         /**
 733          * Get the HTTP method used for this request.
 734          *
 735          * @return string
 736          */
 737         public function getMethod() {
 738                 return $_SERVER['REQUEST_METHOD'] ?? 'GET';
 739         }
 740
 741         /**
 742          * Returns true if the present request was reached by a POST operation,
 743          * false otherwise (GET, HEAD, or command-line).
 744          *
 745          * Note that values retrieved by the object may come from the
 746          * GET URL etc even on a POST request.
 747          *
 748          * @return bool
 749          */
 750         public function wasPosted() {
 751                 return $this->getMethod() == 'POST';
 752         }
 753
 754         /**
 755          * Return the session for this request
 756          *
 757          * This might unpersist an existing session if it was invalid.
 758          *
 759          * @since 1.27
 760          * @note For performance, keep the session locally if you will be making
 761          *  much use of it instead of calling this method repeatedly.
 762          * @return Session
 763          */
 764         public function getSession() {
 765                 if ( $this->sessionId !== null ) {
 766                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 767                         if ( $session ) {
 768                                 return $session;
 769                         }
 770                 }
 771
 772                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 773                 $this->sessionId = $session->getSessionId();
 774                 return $session;
 775         }
 776
 777         /**
 778          * Set the session for this request
 779          * @since 1.27
 780          * @private For use by MediaWiki\Session classes only
 781          * @param SessionId $sessionId
 782          */
 783         public function setSessionId( SessionId $sessionId ) {
 784                 $this->sessionId = $sessionId;
 785         }
 786
 787         /**
 788          * Get the session id for this request, if any
 789          * @since 1.27
 790          * @private For use by MediaWiki\Session classes only
 791          * @return SessionId|null
 792          */
 793         public function getSessionId() {
 794                 return $this->sessionId;
 795         }
 796
 797         /**
 798          * Get a cookie from the $_COOKIE jar
 799          *
 800          * @param string $key The name of the cookie
 801          * @param string|null $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 802          * @param mixed|null $default What to return if the value isn't found
 803          * @return mixed Cookie value or $default if the cookie not set
 804          */
 805         public function getCookie( $key, $prefix = null, $default = null ) {
 806                 if ( $prefix === null ) {
 807                         global $wgCookiePrefix;
 808                         $prefix = $wgCookiePrefix;
 809                 }
 810                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 811         }
 812
 813         /**
 814          * Return the path and query string portion of the main request URI.
 815          * This will be suitable for use as a relative link in HTML output.
 816          *
 817          * @throws MWException
 818          * @return string
 819          */
 820         public static function getGlobalRequestURL() {
 821                 // This method is called on fatal errors; it should not depend on anything complex.
 822
 823                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 824                         $base = $_SERVER['REQUEST_URI'];
 825                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 826                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 827                 ) {
 828                         // Probably IIS; doesn't set REQUEST_URI
 829                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 830                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 831                         $base = $_SERVER['SCRIPT_NAME'];
 832                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 833                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 834                         }
 835                 } else {
 836                         // This shouldn't happen!
 837                         throw new MWException( "Web server doesn't provide either " .
 838                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 839                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 840                 }
 841                 // User-agents should not send a fragment with the URI, but
 842                 // if they do, and the web server passes it on to us, we
 843                 // need to strip it or we get false-positive redirect loops
 844                 // or weird output URLs
 845                 $hash = strpos( $base, '#' );
 846                 if ( $hash !== false ) {
 847                         $base = substr( $base, 0, $hash );
 848                 }
 849
 850                 if ( $base[0] == '/' ) {
 851                         // More than one slash will look like it is protocol relative
 852                         return preg_replace( '!^/+!', '/', $base );
 853                 } else {
 854                         // We may get paths with a host prepended; strip it.
 855                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 856                 }
 857         }
 858
 859         /**
 860          * Return the path and query string portion of the request URI.
 861          * This will be suitable for use as a relative link in HTML output.
 862          *
 863          * @throws MWException
 864          * @return string
 865          */
 866         public function getRequestURL() {
 867                 return self::getGlobalRequestURL();
 868         }
 869
 870         /**
 871          * Return the request URI with the canonical service and hostname, path,
 872          * and query string. This will be suitable for use as an absolute link
 873          * in HTML or other output.
 874          *
 875          * If $wgServer is protocol-relative, this will return a fully
 876          * qualified URL with the protocol of this request object.
 877          *
 878          * @return string
 879          */
 880         public function getFullRequestURL() {
 881                 // Pass an explicit PROTO constant instead of PROTO_CURRENT so that we
 882                 // do not rely on state from the global $wgRequest object (which it would,
 883                 // via wfGetServerUrl/wfExpandUrl/$wgRequest->protocol).
 884                 if ( $this->getProtocol() === 'http' ) {
 885                         return wfGetServerUrl( PROTO_HTTP ) . $this->getRequestURL();
 886                 } else {
 887                         return wfGetServerUrl( PROTO_HTTPS ) . $this->getRequestURL();
 888                 }
 889         }
 890
 891         /**
 892          * @param string $key
 893          * @param string $value
 894          * @return string
 895          */
 896         public function appendQueryValue( $key, $value ) {
 897                 return $this->appendQueryArray( [ $key => $value ] );
 898         }
 899
 900         /**
 901          * Appends or replaces value of query variables.
 902          *
 903          * @param array $array Array of values to replace/add to query
 904          * @return string
 905          */
 906         public function appendQueryArray( $array ) {
 907                 $newquery = $this->getQueryValues();
 908                 unset( $newquery['title'] );
 909                 $newquery = array_merge( $newquery, $array );
 910
 911                 return wfArrayToCgi( $newquery );
 912         }
 913
 914         /**
 915          * Check for limit and offset parameters on the input, and return sensible
 916          * defaults if not given. The limit must be positive and is capped at 5000.
 917          * Offset must be positive but is not capped.
 918          *
 919          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 920          * @param string $optionname To specify an option other than rclimit to pull from.
 921          * @return int[] First element is limit, second is offset
 922          */
 923         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 924                 global $wgUser;
 925
 926                 $limit = $this->getInt( 'limit', 0 );
 927                 if ( $limit < 0 ) {
 928                         $limit = 0;
 929                 }
 930                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 931                         $limit = $wgUser->getIntOption( $optionname );
 932                 }
 933                 if ( $limit <= 0 ) {
 934                         $limit = $deflimit;
 935                 }
 936                 if ( $limit > 5000 ) {
 937                         $limit = 5000; # We have *some* limits...
 938                 }
 939
 940                 $offset = $this->getInt( 'offset', 0 );
 941                 if ( $offset < 0 ) {
 942                         $offset = 0;
 943                 }
 944
 945                 return [ $limit, $offset ];
 946         }
 947
 948         /**
 949          * Return the path to the temporary file where PHP has stored the upload.
 950          *
 951          * @param string $key
 952          * @return string|null String or null if no such file.
 953          */
 954         public function getFileTempname( $key ) {
 955                 $file = new WebRequestUpload( $this, $key );
 956                 return $file->getTempName();
 957         }
 958
 959         /**
 960          * Return the upload error or 0
 961          *
 962          * @param string $key
 963          * @return int
 964          */
 965         public function getUploadError( $key ) {
 966                 $file = new WebRequestUpload( $this, $key );
 967                 return $file->getError();
 968         }
 969
 970         /**
 971          * Return the original filename of the uploaded file, as reported by
 972          * the submitting user agent. HTML-style character entities are
 973          * interpreted and normalized to Unicode normalization form C, in part
 974          * to deal with weird input from Safari with non-ASCII filenames.
 975          *
 976          * Other than this the name is not verified for being a safe filename.
 977          *
 978          * @param string $key
 979          * @return string|null String or null if no such file.
 980          */
 981         public function getFileName( $key ) {
 982                 $file = new WebRequestUpload( $this, $key );
 983                 return $file->getName();
 984         }
 985
 986         /**
 987          * Return a WebRequestUpload object corresponding to the key
 988          *
 989          * @param string $key
 990          * @return WebRequestUpload
 991          */
 992         public function getUpload( $key ) {
 993                 return new WebRequestUpload( $this, $key );
 994         }
 995
 996         /**
 997          * Return a handle to WebResponse style object, for setting cookies,
 998          * headers and other stuff, for Request being worked on.
 999          *
1000          * @return WebResponse
1001          */
1002         public function response() {
1003                 /* Lazy initialization of response object for this request */
1004                 if ( !is_object( $this->response ) ) {
1005                         $class = ( $this instanceof FauxRequest ) ? FauxResponse::class : WebResponse::class;
1006                         $this->response = new $class();
1007                 }
1008                 return $this->response;
1009         }
1010
1011         /**
1012          * Initialise the header list
1013          */
1014         protected function initHeaders() {
1015                 if ( count( $this->headers ) ) {
1016                         return;
1017                 }
1018
1019                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
1020                 if ( $apacheHeaders ) {
1021                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
1022                                 $this->headers[strtoupper( $tempName )] = $tempValue;
1023                         }
1024                 } else {
1025                         foreach ( $_SERVER as $name => $value ) {
1026                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
1027                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
1028                                         $this->headers[$name] = $value;
1029                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
1030                                         $this->headers['CONTENT-LENGTH'] = $value;
1031                                 }
1032                         }
1033                 }
1034         }
1035
1036         /**
1037          * Get an array containing all request headers
1038          *
1039          * @return array Mapping header name to its value
1040          */
1041         public function getAllHeaders() {
1042                 $this->initHeaders();
1043                 return $this->headers;
1044         }
1045
1046         /**
1047          * Get a request header, or false if it isn't set.
1048          *
1049          * @param string $name Case-insensitive header name
1050          * @param int $flags Bitwise combination of:
1051          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
1052          *                               of values, as described in RFC 2616 § 4.2.
1053          *                               (since 1.26).
1054          * @return string|array|bool False if header is unset; otherwise the
1055          *  header value(s) as either a string (the default) or an array, if
1056          *  WebRequest::GETHEADER_LIST flag was set.
1057          */
1058         public function getHeader( $name, $flags = 0 ) {
1059                 $this->initHeaders();
1060                 $name = strtoupper( $name );
1061                 if ( !isset( $this->headers[$name] ) ) {
1062                         return false;
1063                 }
1064                 $value = $this->headers[$name];
1065                 if ( $flags & self::GETHEADER_LIST ) {
1066                         $value = array_map( 'trim', explode( ',', $value ) );
1067                 }
1068                 return $value;
1069         }
1070
1071         /**
1072          * Get data from the session
1073          *
1074          * @note Prefer $this->getSession() instead if making multiple calls.
1075          * @param string $key Name of key in the session
1076          * @return mixed
1077          */
1078         public function getSessionData( $key ) {
1079                 return $this->getSession()->get( $key );
1080         }
1081
1082         /**
1083          * Set session data
1084          *
1085          * @note Prefer $this->getSession() instead if making multiple calls.
1086          * @param string $key Name of key in the session
1087          * @param mixed $data
1088          */
1089         public function setSessionData( $key, $data ) {
1090                 $this->getSession()->set( $key, $data );
1091         }
1092
1093         /**
1094          * Check if Internet Explorer will detect an incorrect cache extension in
1095          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1096          * message or redirect to a safer URL. Returns true if the URL is OK, and
1097          * false if an error message has been shown and the request should be aborted.
1098          *
1099          * @param array $extWhitelist
1100          * @throws HttpError
1101          * @return bool
1102          */
1103         public function checkUrlExtension( $extWhitelist = [] ) {
1104                 $extWhitelist[] = 'php';
1105                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1106                         if ( !$this->wasPosted() ) {
1107                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1108                                         $this->getFullRequestURL(), $extWhitelist );
1109                                 if ( $newUrl !== false ) {
1110                                         $this->doSecurityRedirect( $newUrl );
1111                                         return false;
1112                                 }
1113                         }
1114                         throw new HttpError( 403,
1115                                 'Invalid file extension found in the path info or query string.' );
1116                 }
1117                 return true;
1118         }
1119
1120         /**
1121          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1122          * IE 6. Returns true if it was successful, false otherwise.
1123          *
1124          * @param string $url
1125          * @return bool
1126          */
1127         protected function doSecurityRedirect( $url ) {
1128                 header( 'Location: ' . $url );
1129                 header( 'Content-Type: text/html' );
1130                 $encUrl = htmlspecialchars( $url );
1131                 echo <<<HTML
1132 <!DOCTYPE html>
1133 <html>
1134 <head>
1135 <title>Security redirect</title>
1136 </head>
1137 <body>
1138 <h1>Security redirect</h1>
1139 <p>
1140 We can't serve non-HTML content from the URL you have requested, because
1141 Internet Explorer would interpret it as an incorrect and potentially dangerous
1142 content type.</p>
1143 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1144 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1145 Explorer from seeing a bogus file extension.
1146 </p>
1147 </body>
1148 </html>
1149 HTML;
1150                 echo "\n";
1151                 return true;
1152         }
1153
1154         /**
1155          * Parse the Accept-Language header sent by the client into an array
1156          *
1157          * @return array [ languageCode => q-value ] sorted by q-value in
1158          *   descending order then appearing time in the header in ascending order.
1159          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1160          * This is aligned with rfc2616 section 14.4
1161          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1162          */
1163         public function getAcceptLang() {
1164                 // Modified version of code found at
1165                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1166                 $acceptLang = $this->getHeader( 'Accept-Language' );
1167                 if ( !$acceptLang ) {
1168                         return [];
1169                 }
1170
1171                 // Return the language codes in lower case
1172                 $acceptLang = strtolower( $acceptLang );
1173
1174                 // Break up string into pieces (languages and q factors)
1175                 $lang_parse = null;
1176                 preg_match_all(
1177                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1178                         $acceptLang,
1179                         $lang_parse
1180                 );
1181
1182                 if ( !count( $lang_parse[1] ) ) {
1183                         return [];
1184                 }
1185
1186                 $langcodes = $lang_parse[1];
1187                 $qvalues = $lang_parse[4];
1188                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1189
1190                 // Set default q factor to 1
1191                 foreach ( $indices as $index ) {
1192                         if ( $qvalues[$index] === '' ) {
1193                                 $qvalues[$index] = 1;
1194                         } elseif ( $qvalues[$index] == 0 ) {
1195                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1196                         }
1197                 }
1198
1199                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1200                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1201
1202                 // Create a list like "en" => 0.8
1203                 $langs = array_combine( $langcodes, $qvalues );
1204
1205                 return $langs;
1206         }
1207
1208         /**
1209          * Fetch the raw IP from the request
1210          *
1211          * @since 1.19
1212          *
1213          * @throws MWException
1214          * @return string
1215          */
1216         protected function getRawIP() {
1217                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1218                         return null;
1219                 }
1220
1221                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1222                         throw new MWException( __METHOD__
1223                                 . " : Could not determine the remote IP address due to multiple values." );
1224                 } else {
1225                         $ipchain = $_SERVER['REMOTE_ADDR'];
1226                 }
1227
1228                 return IP::canonicalize( $ipchain );
1229         }
1230
1231         /**
1232          * Work out the IP address based on various globals
1233          * For trusted proxies, use the XFF client IP (first of the chain)
1234          *
1235          * @since 1.19
1236          *
1237          * @throws MWException
1238          * @return string
1239          */
1240         public function getIP() {
1241                 global $wgUsePrivateIPs;
1242
1243                 # Return cached result
1244                 if ( $this->ip !== null ) {
1245                         return $this->ip;
1246                 }
1247
1248                 # collect the originating ips
1249                 $ip = $this->getRawIP();
1250                 if ( !$ip ) {
1251                         throw new MWException( 'Unable to determine IP.' );
1252                 }
1253
1254                 # Append XFF
1255                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1256                 if ( $forwardedFor !== false ) {
1257                         $proxyLookup = MediaWikiServices::getInstance()->getProxyLookup();
1258                         $isConfigured = $proxyLookup->isConfiguredProxy( $ip );
1259                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1260                         $ipchain = array_reverse( $ipchain );
1261                         array_unshift( $ipchain, $ip );
1262
1263                         # Step through XFF list and find the last address in the list which is a
1264                         # trusted server. Set $ip to the IP address given by that trusted server,
1265                         # unless the address is not sensible (e.g. private). However, prefer private
1266                         # IP addresses over proxy servers controlled by this site (more sensible).
1267                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1268                         foreach ( $ipchain as $i => $curIP ) {
1269                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1270                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1271                                         || !$proxyLookup->isTrustedProxy( $curIP )
1272                                 ) {
1273                                         break; // IP is not valid/trusted or does not point to anything
1274                                 }
1275                                 if (
1276                                         IP::isPublic( $ipchain[$i + 1] ) ||
1277                                         $wgUsePrivateIPs ||
1278                                         $proxyLookup->isConfiguredProxy( $curIP ) // T50919; treat IP as sane
1279                                 ) {
1280                                         // Follow the next IP according to the proxy
1281                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1282                                         if ( !$nextIP && $isConfigured ) {
1283                                                 // We have not yet made it past CDN/proxy servers of this site,
1284                                                 // so either they are misconfigured or there is some IP spoofing.
1285                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1286                                         }
1287                                         $ip = $nextIP;
1288                                         // keep traversing the chain
1289                                         continue;
1290                                 }
1291                                 break;
1292                         }
1293                 }
1294
1295                 # Allow extensions to improve our guess
1296                 Hooks::run( 'GetIP', [ &$ip ] );
1297
1298                 if ( !$ip ) {
1299                         throw new MWException( "Unable to determine IP." );
1300                 }
1301
1302                 wfDebug( "IP: $ip\n" );
1303                 $this->ip = $ip;
1304                 return $ip;
1305         }
1306
1307         /**
1308          * @param string $ip
1309          * @return void
1310          * @since 1.21
1311          */
1312         public function setIP( $ip ) {
1313                 $this->ip = $ip;
1314         }
1315
1316         /**
1317          * Check if this request uses a "safe" HTTP method
1318          *
1319          * Safe methods are verbs (e.g. GET/HEAD/OPTIONS) used for obtaining content. Such requests
1320          * are not expected to mutate content, especially in ways attributable to the client. Verbs
1321          * like POST and PUT are typical of non-safe requests which often change content.
1322          *
1323          * @return bool
1324          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1325          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1326          * @since 1.28
1327          */
1328         public function hasSafeMethod() {
1329                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1330                         return false; // CLI mode
1331                 }
1332
1333                 return in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS', 'TRACE' ] );
1334         }
1335
1336         /**
1337          * Whether this request should be identified as being "safe"
1338          *
1339          * This means that the client is not requesting any state changes and that database writes
1340          * are not inherently required. Ideally, no visible updates would happen at all. If they
1341          * must, then they should not be publicly attributed to the end user.
1342          *
1343          * In more detail:
1344          *   - Cache populations and refreshes MAY occur.
1345          *   - Private user session updates and private server logging MAY occur.
1346          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1347          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1348          *
1349          * @return bool
1350          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1351          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1352          * @since 1.28
1353          */
1354         public function isSafeRequest() {
1355                 if ( $this->markedAsSafe && $this->wasPosted() ) {
1356                         return true; // marked as a "safe" POST
1357                 }
1358
1359                 return $this->hasSafeMethod();
1360         }
1361
1362         /**
1363          * Mark this request as identified as being nullipotent even if it is a POST request
1364          *
1365          * POST requests are often used due to the need for a client payload, even if the request
1366          * is otherwise equivalent to a "safe method" request.
1367          *
1368          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1369          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1370          * @since 1.28
1371          */
1372         public function markAsSafeRequest() {
1373                 $this->markedAsSafe = true;
1374         }
1375 }