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                 if ( isset( $arr[$name] ) ) {
 399                         $data = $arr[$name];
 400                         if ( isset( $_GET[$name] ) && is_string( $data ) ) {
 401                                 # Check for alternate/legacy character encoding.
 402                                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 403                                 $data = $contLang->checkTitleEncoding( $data );
 404                         }
 405                         $data = $this->normalizeUnicode( $data );
 406                         return $data;
 407                 } else {
 408                         return $default;
 409                 }
 410         }
 411
 412         /**
 413          * Fetch a scalar from the input without normalization, or return $default
 414          * if it's not set.
 415          *
 416          * Unlike self::getVal(), this does not perform any normalization on the
 417          * input value.
 418          *
 419          * @since 1.28
 420          * @param string $name
 421          * @param string|null $default
 422          * @return string|null
 423          */
 424         public function getRawVal( $name, $default = null ) {
 425                 $name = strtr( $name, '.', '_' ); // See comment in self::getGPCVal()
 426                 if ( isset( $this->data[$name] ) && !is_array( $this->data[$name] ) ) {
 427                         $val = $this->data[$name];
 428                 } else {
 429                         $val = $default;
 430                 }
 431                 if ( is_null( $val ) ) {
 432                         return $val;
 433                 } else {
 434                         return (string)$val;
 435                 }
 436         }
 437
 438         /**
 439          * Fetch a scalar from the input or return $default if it's not set.
 440          * Returns a string. Arrays are discarded. Useful for
 441          * non-freeform text inputs (e.g. predefined internal text keys
 442          * selected by a drop-down menu). For freeform input, see getText().
 443          *
 444          * @param string $name
 445          * @param string|null $default Optional default (or null)
 446          * @return string|null
 447          */
 448         public function getVal( $name, $default = null ) {
 449                 $val = $this->getGPCVal( $this->data, $name, $default );
 450                 if ( is_array( $val ) ) {
 451                         $val = $default;
 452                 }
 453                 if ( is_null( $val ) ) {
 454                         return $val;
 455                 } else {
 456                         return (string)$val;
 457                 }
 458         }
 459
 460         /**
 461          * Set an arbitrary value into our get/post data.
 462          *
 463          * @param string $key Key name to use
 464          * @param mixed $value Value to set
 465          * @return mixed Old value if one was present, null otherwise
 466          */
 467         public function setVal( $key, $value ) {
 468                 $ret = $this->data[$key] ?? null;
 469                 $this->data[$key] = $value;
 470                 return $ret;
 471         }
 472
 473         /**
 474          * Unset an arbitrary value from our get/post data.
 475          *
 476          * @param string $key Key name to use
 477          * @return mixed Old value if one was present, null otherwise
 478          */
 479         public function unsetVal( $key ) {
 480                 if ( !isset( $this->data[$key] ) ) {
 481                         $ret = null;
 482                 } else {
 483                         $ret = $this->data[$key];
 484                         unset( $this->data[$key] );
 485                 }
 486                 return $ret;
 487         }
 488
 489         /**
 490          * Fetch an array from the input or return $default if it's not set.
 491          * If source was scalar, will return an array with a single element.
 492          * If no source and no default, returns null.
 493          *
 494          * @param string $name
 495          * @param array|null $default Optional default (or null)
 496          * @return array|null
 497          */
 498         public function getArray( $name, $default = null ) {
 499                 $val = $this->getGPCVal( $this->data, $name, $default );
 500                 if ( is_null( $val ) ) {
 501                         return null;
 502                 } else {
 503                         return (array)$val;
 504                 }
 505         }
 506
 507         /**
 508          * Fetch an array of integers, or return $default if it's not set.
 509          * If source was scalar, will return an array with a single element.
 510          * If no source and no default, returns null.
 511          * If an array is returned, contents are guaranteed to be integers.
 512          *
 513          * @param string $name
 514          * @param array|null $default Option default (or null)
 515          * @return array Array of ints
 516          */
 517         public function getIntArray( $name, $default = null ) {
 518                 $val = $this->getArray( $name, $default );
 519                 if ( is_array( $val ) ) {
 520                         $val = array_map( 'intval', $val );
 521                 }
 522                 return $val;
 523         }
 524
 525         /**
 526          * Fetch an integer value from the input or return $default if not set.
 527          * Guaranteed to return an integer; non-numeric input will typically
 528          * return 0.
 529          *
 530          * @param string $name
 531          * @param int $default
 532          * @return int
 533          */
 534         public function getInt( $name, $default = 0 ) {
 535                 return intval( $this->getRawVal( $name, $default ) );
 536         }
 537
 538         /**
 539          * Fetch an integer value from the input or return null if empty.
 540          * Guaranteed to return an integer or null; non-numeric input will
 541          * typically return null.
 542          *
 543          * @param string $name
 544          * @return int|null
 545          */
 546         public function getIntOrNull( $name ) {
 547                 $val = $this->getRawVal( $name );
 548                 return is_numeric( $val )
 549                         ? intval( $val )
 550                         : null;
 551         }
 552
 553         /**
 554          * Fetch a floating point value from the input or return $default if not set.
 555          * Guaranteed to return a float; non-numeric input will typically
 556          * return 0.
 557          *
 558          * @since 1.23
 559          * @param string $name
 560          * @param float $default
 561          * @return float
 562          */
 563         public function getFloat( $name, $default = 0.0 ) {
 564                 return floatval( $this->getRawVal( $name, $default ) );
 565         }
 566
 567         /**
 568          * Fetch a boolean value from the input or return $default if not set.
 569          * Guaranteed to return true or false, with normal PHP semantics for
 570          * boolean interpretation of strings.
 571          *
 572          * @param string $name
 573          * @param bool $default
 574          * @return bool
 575          */
 576         public function getBool( $name, $default = false ) {
 577                 return (bool)$this->getRawVal( $name, $default );
 578         }
 579
 580         /**
 581          * Fetch a boolean value from the input or return $default if not set.
 582          * Unlike getBool, the string "false" will result in boolean false, which is
 583          * useful when interpreting information sent from JavaScript.
 584          *
 585          * @param string $name
 586          * @param bool $default
 587          * @return bool
 588          */
 589         public function getFuzzyBool( $name, $default = false ) {
 590                 return $this->getBool( $name, $default )
 591                         && strcasecmp( $this->getRawVal( $name ), 'false' ) !== 0;
 592         }
 593
 594         /**
 595          * Return true if the named value is set in the input, whatever that
 596          * value is (even "0"). Return false if the named value is not set.
 597          * Example use is checking for the presence of check boxes in forms.
 598          *
 599          * @param string $name
 600          * @return bool
 601          */
 602         public function getCheck( $name ) {
 603                 # Checkboxes and buttons are only present when clicked
 604                 # Presence connotes truth, absence false
 605                 return $this->getRawVal( $name, null ) !== null;
 606         }
 607
 608         /**
 609          * Fetch a text string from the given array or return $default if it's not
 610          * set. Carriage returns are stripped from the text. This should generally
 611          * be used for form "<textarea>" and "<input>" fields, and for
 612          * user-supplied freeform text input.
 613          *
 614          * @param string $name
 615          * @param string $default Optional
 616          * @return string
 617          */
 618         public function getText( $name, $default = '' ) {
 619                 $val = $this->getVal( $name, $default );
 620                 return str_replace( "\r\n", "\n", $val );
 621         }
 622
 623         /**
 624          * Extracts the given named values into an array.
 625          * If no arguments are given, returns all input values.
 626          * No transformation is performed on the values.
 627          *
 628          * @return array
 629          */
 630         public function getValues() {
 631                 $names = func_get_args();
 632                 if ( count( $names ) == 0 ) {
 633                         $names = array_keys( $this->data );
 634                 }
 635
 636                 $retVal = [];
 637                 foreach ( $names as $name ) {
 638                         $value = $this->getGPCVal( $this->data, $name, null );
 639                         if ( !is_null( $value ) ) {
 640                                 $retVal[$name] = $value;
 641                         }
 642                 }
 643                 return $retVal;
 644         }
 645
 646         /**
 647          * Returns the names of all input values excluding those in $exclude.
 648          *
 649          * @param array $exclude
 650          * @return array
 651          */
 652         public function getValueNames( $exclude = [] ) {
 653                 return array_diff( array_keys( $this->getValues() ), $exclude );
 654         }
 655
 656         /**
 657          * Get the values passed in the query string.
 658          * No transformation is performed on the values.
 659          *
 660          * @codeCoverageIgnore
 661          * @return array
 662          */
 663         public function getQueryValues() {
 664                 return $_GET;
 665         }
 666
 667         /**
 668          * Get the values passed via POST.
 669          * No transformation is performed on the values.
 670          *
 671          * @since 1.32
 672          * @codeCoverageIgnore
 673          * @return array
 674          */
 675         public function getPostValues() {
 676                 return $_POST;
 677         }
 678
 679         /**
 680          * Return the contents of the Query with no decoding. Use when you need to
 681          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 682          *
 683          * @codeCoverageIgnore
 684          * @return string
 685          */
 686         public function getRawQueryString() {
 687                 return $_SERVER['QUERY_STRING'];
 688         }
 689
 690         /**
 691          * Return the contents of the POST with no decoding. Use when you need to
 692          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 693          *
 694          * @return string
 695          */
 696         public function getRawPostString() {
 697                 if ( !$this->wasPosted() ) {
 698                         return '';
 699                 }
 700                 return $this->getRawInput();
 701         }
 702
 703         /**
 704          * Return the raw request body, with no processing. Cached since some methods
 705          * disallow reading the stream more than once. As stated in the php docs, this
 706          * does not work with enctype="multipart/form-data".
 707          *
 708          * @return string
 709          */
 710         public function getRawInput() {
 711                 static $input = null;
 712                 if ( $input === null ) {
 713                         $input = file_get_contents( 'php://input' );
 714                 }
 715                 return $input;
 716         }
 717
 718         /**
 719          * Get the HTTP method used for this request.
 720          *
 721          * @return string
 722          */
 723         public function getMethod() {
 724                 return $_SERVER['REQUEST_METHOD'] ?? 'GET';
 725         }
 726
 727         /**
 728          * Returns true if the present request was reached by a POST operation,
 729          * false otherwise (GET, HEAD, or command-line).
 730          *
 731          * Note that values retrieved by the object may come from the
 732          * GET URL etc even on a POST request.
 733          *
 734          * @return bool
 735          */
 736         public function wasPosted() {
 737                 return $this->getMethod() == 'POST';
 738         }
 739
 740         /**
 741          * Return the session for this request
 742          *
 743          * This might unpersist an existing session if it was invalid.
 744          *
 745          * @since 1.27
 746          * @note For performance, keep the session locally if you will be making
 747          *  much use of it instead of calling this method repeatedly.
 748          * @return Session
 749          */
 750         public function getSession() {
 751                 if ( $this->sessionId !== null ) {
 752                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 753                         if ( $session ) {
 754                                 return $session;
 755                         }
 756                 }
 757
 758                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 759                 $this->sessionId = $session->getSessionId();
 760                 return $session;
 761         }
 762
 763         /**
 764          * Set the session for this request
 765          * @since 1.27
 766          * @private For use by MediaWiki\Session classes only
 767          * @param SessionId $sessionId
 768          */
 769         public function setSessionId( SessionId $sessionId ) {
 770                 $this->sessionId = $sessionId;
 771         }
 772
 773         /**
 774          * Get the session id for this request, if any
 775          * @since 1.27
 776          * @private For use by MediaWiki\Session classes only
 777          * @return SessionId|null
 778          */
 779         public function getSessionId() {
 780                 return $this->sessionId;
 781         }
 782
 783         /**
 784          * Get a cookie from the $_COOKIE jar
 785          *
 786          * @param string $key The name of the cookie
 787          * @param string|null $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 788          * @param mixed|null $default What to return if the value isn't found
 789          * @return mixed Cookie value or $default if the cookie not set
 790          */
 791         public function getCookie( $key, $prefix = null, $default = null ) {
 792                 if ( $prefix === null ) {
 793                         global $wgCookiePrefix;
 794                         $prefix = $wgCookiePrefix;
 795                 }
 796                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 797         }
 798
 799         /**
 800          * Return the path and query string portion of the main request URI.
 801          * This will be suitable for use as a relative link in HTML output.
 802          *
 803          * @throws MWException
 804          * @return string
 805          */
 806         public static function getGlobalRequestURL() {
 807                 // This method is called on fatal errors; it should not depend on anything complex.
 808
 809                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 810                         $base = $_SERVER['REQUEST_URI'];
 811                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 812                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 813                 ) {
 814                         // Probably IIS; doesn't set REQUEST_URI
 815                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 816                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 817                         $base = $_SERVER['SCRIPT_NAME'];
 818                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 819                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 820                         }
 821                 } else {
 822                         // This shouldn't happen!
 823                         throw new MWException( "Web server doesn't provide either " .
 824                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 825                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 826                 }
 827                 // User-agents should not send a fragment with the URI, but
 828                 // if they do, and the web server passes it on to us, we
 829                 // need to strip it or we get false-positive redirect loops
 830                 // or weird output URLs
 831                 $hash = strpos( $base, '#' );
 832                 if ( $hash !== false ) {
 833                         $base = substr( $base, 0, $hash );
 834                 }
 835
 836                 if ( $base[0] == '/' ) {
 837                         // More than one slash will look like it is protocol relative
 838                         return preg_replace( '!^/+!', '/', $base );
 839                 } else {
 840                         // We may get paths with a host prepended; strip it.
 841                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 842                 }
 843         }
 844
 845         /**
 846          * Return the path and query string portion of the request URI.
 847          * This will be suitable for use as a relative link in HTML output.
 848          *
 849          * @throws MWException
 850          * @return string
 851          */
 852         public function getRequestURL() {
 853                 return self::getGlobalRequestURL();
 854         }
 855
 856         /**
 857          * Return the request URI with the canonical service and hostname, path,
 858          * and query string. This will be suitable for use as an absolute link
 859          * in HTML or other output.
 860          *
 861          * If $wgServer is protocol-relative, this will return a fully
 862          * qualified URL with the protocol of this request object.
 863          *
 864          * @return string
 865          */
 866         public function getFullRequestURL() {
 867                 // Pass an explicit PROTO constant instead of PROTO_CURRENT so that we
 868                 // do not rely on state from the global $wgRequest object (which it would,
 869                 // via wfGetServerUrl/wfExpandUrl/$wgRequest->protocol).
 870                 if ( $this->getProtocol() === 'http' ) {
 871                         return wfGetServerUrl( PROTO_HTTP ) . $this->getRequestURL();
 872                 } else {
 873                         return wfGetServerUrl( PROTO_HTTPS ) . $this->getRequestURL();
 874                 }
 875         }
 876
 877         /**
 878          * @param string $key
 879          * @param string $value
 880          * @return string
 881          */
 882         public function appendQueryValue( $key, $value ) {
 883                 return $this->appendQueryArray( [ $key => $value ] );
 884         }
 885
 886         /**
 887          * Appends or replaces value of query variables.
 888          *
 889          * @param array $array Array of values to replace/add to query
 890          * @return string
 891          */
 892         public function appendQueryArray( $array ) {
 893                 $newquery = $this->getQueryValues();
 894                 unset( $newquery['title'] );
 895                 $newquery = array_merge( $newquery, $array );
 896
 897                 return wfArrayToCgi( $newquery );
 898         }
 899
 900         /**
 901          * Check for limit and offset parameters on the input, and return sensible
 902          * defaults if not given. The limit must be positive and is capped at 5000.
 903          * Offset must be positive but is not capped.
 904          *
 905          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 906          * @param string $optionname To specify an option other than rclimit to pull from.
 907          * @return int[] First element is limit, second is offset
 908          */
 909         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 910                 global $wgUser;
 911
 912                 $limit = $this->getInt( 'limit', 0 );
 913                 if ( $limit < 0 ) {
 914                         $limit = 0;
 915                 }
 916                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 917                         $limit = $wgUser->getIntOption( $optionname );
 918                 }
 919                 if ( $limit <= 0 ) {
 920                         $limit = $deflimit;
 921                 }
 922                 if ( $limit > 5000 ) {
 923                         $limit = 5000; # We have *some* limits...
 924                 }
 925
 926                 $offset = $this->getInt( 'offset', 0 );
 927                 if ( $offset < 0 ) {
 928                         $offset = 0;
 929                 }
 930
 931                 return [ $limit, $offset ];
 932         }
 933
 934         /**
 935          * Return the path to the temporary file where PHP has stored the upload.
 936          *
 937          * @param string $key
 938          * @return string|null String or null if no such file.
 939          */
 940         public function getFileTempname( $key ) {
 941                 $file = new WebRequestUpload( $this, $key );
 942                 return $file->getTempName();
 943         }
 944
 945         /**
 946          * Return the upload error or 0
 947          *
 948          * @param string $key
 949          * @return int
 950          */
 951         public function getUploadError( $key ) {
 952                 $file = new WebRequestUpload( $this, $key );
 953                 return $file->getError();
 954         }
 955
 956         /**
 957          * Return the original filename of the uploaded file, as reported by
 958          * the submitting user agent. HTML-style character entities are
 959          * interpreted and normalized to Unicode normalization form C, in part
 960          * to deal with weird input from Safari with non-ASCII filenames.
 961          *
 962          * Other than this the name is not verified for being a safe filename.
 963          *
 964          * @param string $key
 965          * @return string|null String or null if no such file.
 966          */
 967         public function getFileName( $key ) {
 968                 $file = new WebRequestUpload( $this, $key );
 969                 return $file->getName();
 970         }
 971
 972         /**
 973          * Return a WebRequestUpload object corresponding to the key
 974          *
 975          * @param string $key
 976          * @return WebRequestUpload
 977          */
 978         public function getUpload( $key ) {
 979                 return new WebRequestUpload( $this, $key );
 980         }
 981
 982         /**
 983          * Return a handle to WebResponse style object, for setting cookies,
 984          * headers and other stuff, for Request being worked on.
 985          *
 986          * @return WebResponse
 987          */
 988         public function response() {
 989                 /* Lazy initialization of response object for this request */
 990                 if ( !is_object( $this->response ) ) {
 991                         $class = ( $this instanceof FauxRequest ) ? FauxResponse::class : WebResponse::class;
 992                         $this->response = new $class();
 993                 }
 994                 return $this->response;
 995         }
 996
 997         /**
 998          * Initialise the header list
 999          */
1000         protected function initHeaders() {
1001                 if ( count( $this->headers ) ) {
1002                         return;
1003                 }
1004
1005                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
1006                 if ( $apacheHeaders ) {
1007                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
1008                                 $this->headers[strtoupper( $tempName )] = $tempValue;
1009                         }
1010                 } else {
1011                         foreach ( $_SERVER as $name => $value ) {
1012                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
1013                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
1014                                         $this->headers[$name] = $value;
1015                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
1016                                         $this->headers['CONTENT-LENGTH'] = $value;
1017                                 }
1018                         }
1019                 }
1020         }
1021
1022         /**
1023          * Get an array containing all request headers
1024          *
1025          * @return array Mapping header name to its value
1026          */
1027         public function getAllHeaders() {
1028                 $this->initHeaders();
1029                 return $this->headers;
1030         }
1031
1032         /**
1033          * Get a request header, or false if it isn't set.
1034          *
1035          * @param string $name Case-insensitive header name
1036          * @param int $flags Bitwise combination of:
1037          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
1038          *                               of values, as described in RFC 2616 § 4.2.
1039          *                               (since 1.26).
1040          * @return string|array|bool False if header is unset; otherwise the
1041          *  header value(s) as either a string (the default) or an array, if
1042          *  WebRequest::GETHEADER_LIST flag was set.
1043          */
1044         public function getHeader( $name, $flags = 0 ) {
1045                 $this->initHeaders();
1046                 $name = strtoupper( $name );
1047                 if ( !isset( $this->headers[$name] ) ) {
1048                         return false;
1049                 }
1050                 $value = $this->headers[$name];
1051                 if ( $flags & self::GETHEADER_LIST ) {
1052                         $value = array_map( 'trim', explode( ',', $value ) );
1053                 }
1054                 return $value;
1055         }
1056
1057         /**
1058          * Get data from the session
1059          *
1060          * @note Prefer $this->getSession() instead if making multiple calls.
1061          * @param string $key Name of key in the session
1062          * @return mixed
1063          */
1064         public function getSessionData( $key ) {
1065                 return $this->getSession()->get( $key );
1066         }
1067
1068         /**
1069          * Set session data
1070          *
1071          * @note Prefer $this->getSession() instead if making multiple calls.
1072          * @param string $key Name of key in the session
1073          * @param mixed $data
1074          */
1075         public function setSessionData( $key, $data ) {
1076                 $this->getSession()->set( $key, $data );
1077         }
1078
1079         /**
1080          * Check if Internet Explorer will detect an incorrect cache extension in
1081          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1082          * message or redirect to a safer URL. Returns true if the URL is OK, and
1083          * false if an error message has been shown and the request should be aborted.
1084          *
1085          * @param array $extWhitelist
1086          * @throws HttpError
1087          * @return bool
1088          */
1089         public function checkUrlExtension( $extWhitelist = [] ) {
1090                 $extWhitelist[] = 'php';
1091                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1092                         if ( !$this->wasPosted() ) {
1093                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1094                                         $this->getFullRequestURL(), $extWhitelist );
1095                                 if ( $newUrl !== false ) {
1096                                         $this->doSecurityRedirect( $newUrl );
1097                                         return false;
1098                                 }
1099                         }
1100                         throw new HttpError( 403,
1101                                 'Invalid file extension found in the path info or query string.' );
1102                 }
1103                 return true;
1104         }
1105
1106         /**
1107          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1108          * IE 6. Returns true if it was successful, false otherwise.
1109          *
1110          * @param string $url
1111          * @return bool
1112          */
1113         protected function doSecurityRedirect( $url ) {
1114                 header( 'Location: ' . $url );
1115                 header( 'Content-Type: text/html' );
1116                 $encUrl = htmlspecialchars( $url );
1117                 echo <<<HTML
1118 <!DOCTYPE html>
1119 <html>
1120 <head>
1121 <title>Security redirect</title>
1122 </head>
1123 <body>
1124 <h1>Security redirect</h1>
1125 <p>
1126 We can't serve non-HTML content from the URL you have requested, because
1127 Internet Explorer would interpret it as an incorrect and potentially dangerous
1128 content type.</p>
1129 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1130 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1131 Explorer from seeing a bogus file extension.
1132 </p>
1133 </body>
1134 </html>
1135 HTML;
1136                 echo "\n";
1137                 return true;
1138         }
1139
1140         /**
1141          * Parse the Accept-Language header sent by the client into an array
1142          *
1143          * @return array [ languageCode => q-value ] sorted by q-value in
1144          *   descending order then appearing time in the header in ascending order.
1145          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1146          * This is aligned with rfc2616 section 14.4
1147          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1148          */
1149         public function getAcceptLang() {
1150                 // Modified version of code found at
1151                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1152                 $acceptLang = $this->getHeader( 'Accept-Language' );
1153                 if ( !$acceptLang ) {
1154                         return [];
1155                 }
1156
1157                 // Return the language codes in lower case
1158                 $acceptLang = strtolower( $acceptLang );
1159
1160                 // Break up string into pieces (languages and q factors)
1161                 $lang_parse = null;
1162                 preg_match_all(
1163                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1164                         $acceptLang,
1165                         $lang_parse
1166                 );
1167
1168                 if ( !count( $lang_parse[1] ) ) {
1169                         return [];
1170                 }
1171
1172                 $langcodes = $lang_parse[1];
1173                 $qvalues = $lang_parse[4];
1174                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1175
1176                 // Set default q factor to 1
1177                 foreach ( $indices as $index ) {
1178                         if ( $qvalues[$index] === '' ) {
1179                                 $qvalues[$index] = 1;
1180                         } elseif ( $qvalues[$index] == 0 ) {
1181                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1182                         }
1183                 }
1184
1185                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1186                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1187
1188                 // Create a list like "en" => 0.8
1189                 $langs = array_combine( $langcodes, $qvalues );
1190
1191                 return $langs;
1192         }
1193
1194         /**
1195          * Fetch the raw IP from the request
1196          *
1197          * @since 1.19
1198          *
1199          * @throws MWException
1200          * @return string
1201          */
1202         protected function getRawIP() {
1203                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1204                         return null;
1205                 }
1206
1207                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1208                         throw new MWException( __METHOD__
1209                                 . " : Could not determine the remote IP address due to multiple values." );
1210                 } else {
1211                         $ipchain = $_SERVER['REMOTE_ADDR'];
1212                 }
1213
1214                 return IP::canonicalize( $ipchain );
1215         }
1216
1217         /**
1218          * Work out the IP address based on various globals
1219          * For trusted proxies, use the XFF client IP (first of the chain)
1220          *
1221          * @since 1.19
1222          *
1223          * @throws MWException
1224          * @return string
1225          */
1226         public function getIP() {
1227                 global $wgUsePrivateIPs;
1228
1229                 # Return cached result
1230                 if ( $this->ip !== null ) {
1231                         return $this->ip;
1232                 }
1233
1234                 # collect the originating ips
1235                 $ip = $this->getRawIP();
1236                 if ( !$ip ) {
1237                         throw new MWException( 'Unable to determine IP.' );
1238                 }
1239
1240                 # Append XFF
1241                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1242                 if ( $forwardedFor !== false ) {
1243                         $proxyLookup = MediaWikiServices::getInstance()->getProxyLookup();
1244                         $isConfigured = $proxyLookup->isConfiguredProxy( $ip );
1245                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1246                         $ipchain = array_reverse( $ipchain );
1247                         array_unshift( $ipchain, $ip );
1248
1249                         # Step through XFF list and find the last address in the list which is a
1250                         # trusted server. Set $ip to the IP address given by that trusted server,
1251                         # unless the address is not sensible (e.g. private). However, prefer private
1252                         # IP addresses over proxy servers controlled by this site (more sensible).
1253                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1254                         foreach ( $ipchain as $i => $curIP ) {
1255                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1256                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1257                                         || !$proxyLookup->isTrustedProxy( $curIP )
1258                                 ) {
1259                                         break; // IP is not valid/trusted or does not point to anything
1260                                 }
1261                                 if (
1262                                         IP::isPublic( $ipchain[$i + 1] ) ||
1263                                         $wgUsePrivateIPs ||
1264                                         $proxyLookup->isConfiguredProxy( $curIP ) // T50919; treat IP as sane
1265                                 ) {
1266                                         // Follow the next IP according to the proxy
1267                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1268                                         if ( !$nextIP && $isConfigured ) {
1269                                                 // We have not yet made it past CDN/proxy servers of this site,
1270                                                 // so either they are misconfigured or there is some IP spoofing.
1271                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1272                                         }
1273                                         $ip = $nextIP;
1274                                         // keep traversing the chain
1275                                         continue;
1276                                 }
1277                                 break;
1278                         }
1279                 }
1280
1281                 # Allow extensions to improve our guess
1282                 Hooks::run( 'GetIP', [ &$ip ] );
1283
1284                 if ( !$ip ) {
1285                         throw new MWException( "Unable to determine IP." );
1286                 }
1287
1288                 wfDebug( "IP: $ip\n" );
1289                 $this->ip = $ip;
1290                 return $ip;
1291         }
1292
1293         /**
1294          * @param string $ip
1295          * @return void
1296          * @since 1.21
1297          */
1298         public function setIP( $ip ) {
1299                 $this->ip = $ip;
1300         }
1301
1302         /**
1303          * Check if this request uses a "safe" HTTP method
1304          *
1305          * Safe methods are verbs (e.g. GET/HEAD/OPTIONS) used for obtaining content. Such requests
1306          * are not expected to mutate content, especially in ways attributable to the client. Verbs
1307          * like POST and PUT are typical of non-safe requests which often change content.
1308          *
1309          * @return bool
1310          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1311          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1312          * @since 1.28
1313          */
1314         public function hasSafeMethod() {
1315                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1316                         return false; // CLI mode
1317                 }
1318
1319                 return in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS', 'TRACE' ] );
1320         }
1321
1322         /**
1323          * Whether this request should be identified as being "safe"
1324          *
1325          * This means that the client is not requesting any state changes and that database writes
1326          * are not inherently required. Ideally, no visible updates would happen at all. If they
1327          * must, then they should not be publicly attributed to the end user.
1328          *
1329          * In more detail:
1330          *   - Cache populations and refreshes MAY occur.
1331          *   - Private user session updates and private server logging MAY occur.
1332          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1333          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1334          *
1335          * @return bool
1336          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1337          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1338          * @since 1.28
1339          */
1340         public function isSafeRequest() {
1341                 if ( $this->markedAsSafe && $this->wasPosted() ) {
1342                         return true; // marked as a "safe" POST
1343                 }
1344
1345                 return $this->hasSafeMethod();
1346         }
1347
1348         /**
1349          * Mark this request as identified as being nullipotent even if it is a POST request
1350          *
1351          * POST requests are often used due to the need for a client payload, even if the request
1352          * is otherwise equivalent to a "safe method" request.
1353          *
1354          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1355          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1356          * @since 1.28
1357          */
1358         public function markAsSafeRequest() {
1359                 $this->markedAsSafe = true;
1360         }
1361 }