includes/WebRequest.php

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