includes/WebRequest.php

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