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