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