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 /**
  32  * The WebRequest class encapsulates getting at data passed in the
  33  * URL or via a POSTed form stripping illegal input characters and
  34  * normalizing Unicode sequences.
  35  *
  36  * @ingroup HTTP
  37  */
  38 class WebRequest {
  39         protected $data, $headers = [];
  40
  41         /**
  42          * Flag to make WebRequest::getHeader return an array of values.
  43          * @since 1.26
  44          */
  45         const GETHEADER_LIST = 1;
  46
  47         /**
  48          * The unique request ID.
  49          * @var string
  50          */
  51         private static $reqId;
  52
  53         /**
  54          * Lazy-init response object
  55          * @var WebResponse
  56          */
  57         private $response;
  58
  59         /**
  60          * Cached client IP address
  61          * @var string
  62          */
  63         private $ip;
  64
  65         /**
  66          * The timestamp of the start of the request, with microsecond precision.
  67          * @var float
  68          */
  69         protected $requestTime;
  70
  71         /**
  72          * Cached URL protocol
  73          * @var string
  74          */
  75         protected $protocol;
  76
  77         /**
  78          * @var SessionId|null Session ID to use for this
  79          *  request. We can't save the session directly due to reference cycles not
  80          *  working too well (slow GC in Zend and never collected in HHVM).
  81          */
  82         protected $sessionId = null;
  83
  84         /** @var bool Whether this HTTP request is "safe" (even if it is an HTTP post) */
  85         protected $markedAsSafe = false;
  86
  87         /**
  88          * @codeCoverageIgnore
  89          */
  90         public function __construct() {
  91                 $this->requestTime = isset( $_SERVER['REQUEST_TIME_FLOAT'] )
  92                         ? $_SERVER['REQUEST_TIME_FLOAT'] : microtime( true );
  93
  94                 // POST overrides GET data
  95                 // We don't use $_REQUEST here to avoid interference from cookies...
  96                 $this->data = $_POST + $_GET;
  97         }
  98
  99         /**
 100          * Extract relevant query arguments from the http request uri's path
 101          * to be merged with the normal php provided query arguments.
 102          * Tries to use the REQUEST_URI data if available and parses it
 103          * according to the wiki's configuration looking for any known pattern.
 104          *
 105          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
 106          * provided by the server if any and use that to set a 'title' parameter.
 107          *
 108          * @param string $want If this is not 'all', then the function
 109          * will return an empty array if it determines that the URL is
 110          * inside a rewrite path.
 111          *
 112          * @return array Any query arguments found in path matches.
 113          */
 114         public static function getPathInfo( $want = 'all' ) {
 115                 global $wgUsePathInfo;
 116                 // PATH_INFO is mangled due to https://bugs.php.net/bug.php?id=31892
 117                 // And also by Apache 2.x, double slashes are converted to single slashes.
 118                 // So we will use REQUEST_URI if possible.
 119                 $matches = [];
 120                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
 121                         // Slurp out the path portion to examine...
 122                         $url = $_SERVER['REQUEST_URI'];
 123                         if ( !preg_match( '!^https?://!', $url ) ) {
 124                                 $url = 'http://unused' . $url;
 125                         }
 126                         MediaWiki\suppressWarnings();
 127                         $a = parse_url( $url );
 128                         MediaWiki\restoreWarnings();
 129                         if ( $a ) {
 130                                 $path = isset( $a['path'] ) ? $a['path'] : '';
 131
 132                                 global $wgScript;
 133                                 if ( $path == $wgScript && $want !== 'all' ) {
 134                                         // Script inside a rewrite path?
 135                                         // Abort to keep from breaking...
 136                                         return $matches;
 137                                 }
 138
 139                                 $router = new PathRouter;
 140
 141                                 // Raw PATH_INFO style
 142                                 $router->add( "$wgScript/$1" );
 143
 144                                 if ( isset( $_SERVER['SCRIPT_NAME'] )
 145                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] )
 146                                 ) {
 147                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 148                                         # But we do have some other .php files such as img_auth.php
 149                                         # Don't let root article paths clober the parsing for them
 150                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 151                                 }
 152
 153                                 global $wgArticlePath;
 154                                 if ( $wgArticlePath ) {
 155                                         $router->add( $wgArticlePath );
 156                                 }
 157
 158                                 global $wgActionPaths;
 159                                 if ( $wgActionPaths ) {
 160                                         $router->add( $wgActionPaths, [ 'action' => '$key' ] );
 161                                 }
 162
 163                                 global $wgVariantArticlePath, $wgContLang;
 164                                 if ( $wgVariantArticlePath ) {
 165                                         $router->add( $wgVariantArticlePath,
 166                                                 [ 'variant' => '$2' ],
 167                                                 [ '$2' => $wgContLang->getVariants() ]
 168                                         );
 169                                 }
 170
 171                                 Hooks::run( 'WebRequestPathInfoRouter', [ $router ] );
 172
 173                                 $matches = $router->parse( $path );
 174                         }
 175                 } elseif ( $wgUsePathInfo ) {
 176                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 177                                 // Mangled PATH_INFO
 178                                 // https://bugs.php.net/bug.php?id=31892
 179                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 180                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 181
 182                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && $_SERVER['PATH_INFO'] != '' ) {
 183                                 // Regular old PATH_INFO yay
 184                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 185                         }
 186                 }
 187
 188                 return $matches;
 189         }
 190
 191         /**
 192          * Work out an appropriate URL prefix containing scheme and host, based on
 193          * information detected from $_SERVER
 194          *
 195          * @return string
 196          */
 197         public static function detectServer() {
 198                 global $wgAssumeProxiesUseDefaultProtocolPorts;
 199
 200                 $proto = self::detectProtocol();
 201                 $stdPort = $proto === 'https' ? 443 : 80;
 202
 203                 $varNames = [ 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' ];
 204                 $host = 'localhost';
 205                 $port = $stdPort;
 206                 foreach ( $varNames as $varName ) {
 207                         if ( !isset( $_SERVER[$varName] ) ) {
 208                                 continue;
 209                         }
 210
 211                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 212                         if ( !$parts ) {
 213                                 // Invalid, do not use
 214                                 continue;
 215                         }
 216
 217                         $host = $parts[0];
 218                         if ( $wgAssumeProxiesUseDefaultProtocolPorts && isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) ) {
 219                                 // T72021: Assume that upstream proxy is running on the default
 220                                 // port based on the protocol. We have no reliable way to determine
 221                                 // the actual port in use upstream.
 222                                 $port = $stdPort;
 223                         } elseif ( $parts[1] === false ) {
 224                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 225                                         $port = $_SERVER['SERVER_PORT'];
 226                                 } // else leave it as $stdPort
 227                         } else {
 228                                 $port = $parts[1];
 229                         }
 230                         break;
 231                 }
 232
 233                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 234         }
 235
 236         /**
 237          * Detect the protocol from $_SERVER.
 238          * This is for use prior to Setup.php, when no WebRequest object is available.
 239          * At other times, use the non-static function getProtocol().
 240          *
 241          * @return string
 242          */
 243         public static function detectProtocol() {
 244                 if ( ( !empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) ||
 245                         ( isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) &&
 246                         $_SERVER['HTTP_X_FORWARDED_PROTO'] === 'https' ) ) {
 247                         return 'https';
 248                 } else {
 249                         return 'http';
 250                 }
 251         }
 252
 253         /**
 254          * Get the number of seconds to have elapsed since request start,
 255          * in fractional seconds, with microsecond resolution.
 256          *
 257          * @return float
 258          * @since 1.25
 259          */
 260         public function getElapsedTime() {
 261                 return microtime( true ) - $this->requestTime;
 262         }
 263
 264         /**
 265          * Get the unique request ID.
 266          * This is either the value of the UNIQUE_ID envvar (if present) or a
 267          * randomly-generated 24-character string.
 268          *
 269          * @return string
 270          * @since 1.27
 271          */
 272         public static function getRequestId() {
 273                 if ( !self::$reqId ) {
 274                         self::$reqId = isset( $_SERVER['UNIQUE_ID'] )
 275                                 ? $_SERVER['UNIQUE_ID'] : wfRandomString( 24 );
 276                 }
 277
 278                 return self::$reqId;
 279         }
 280
 281         /**
 282          * Override the unique request ID. This is for sub-requests, such as jobs,
 283          * that wish to use the same id but are not part of the same execution context.
 284          *
 285          * @param string $id
 286          * @since 1.27
 287          */
 288         public static function overrideRequestId( $id ) {
 289                 self::$reqId = $id;
 290         }
 291
 292         /**
 293          * Get the current URL protocol (http or https)
 294          * @return string
 295          */
 296         public function getProtocol() {
 297                 if ( $this->protocol === null ) {
 298                         $this->protocol = self::detectProtocol();
 299                 }
 300                 return $this->protocol;
 301         }
 302
 303         /**
 304          * Check for title, action, and/or variant data in the URL
 305          * and interpolate it into the GET variables.
 306          * This should only be run after $wgContLang is available,
 307          * as we may need the list of language variants to determine
 308          * available variant URLs.
 309          */
 310         public function interpolateTitle() {
 311                 // T18019: title interpolation on API queries is useless and sometimes harmful
 312                 if ( defined( 'MW_API' ) ) {
 313                         return;
 314                 }
 315
 316                 $matches = self::getPathInfo( 'title' );
 317                 foreach ( $matches as $key => $val ) {
 318                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 319                 }
 320         }
 321
 322         /**
 323          * URL rewriting function; tries to extract page title and,
 324          * optionally, one other fixed parameter value from a URL path.
 325          *
 326          * @param string $path The URL path given from the client
 327          * @param array $bases One or more URLs, optionally with $1 at the end
 328          * @param string|bool $key If provided, the matching key in $bases will be
 329          *    passed on as the value of this URL parameter
 330          * @return array Array of URL variables to interpolate; empty if no match
 331          */
 332         static function extractTitle( $path, $bases, $key = false ) {
 333                 foreach ( (array)$bases as $keyValue => $base ) {
 334                         // Find the part after $wgArticlePath
 335                         $base = str_replace( '$1', '', $base );
 336                         $baseLen = strlen( $base );
 337                         if ( substr( $path, 0, $baseLen ) == $base ) {
 338                                 $raw = substr( $path, $baseLen );
 339                                 if ( $raw !== '' ) {
 340                                         $matches = [ 'title' => rawurldecode( $raw ) ];
 341                                         if ( $key ) {
 342                                                 $matches[$key] = $keyValue;
 343                                         }
 344                                         return $matches;
 345                                 }
 346                         }
 347                 }
 348                 return [];
 349         }
 350
 351         /**
 352          * Recursively normalizes UTF-8 strings in the given array.
 353          *
 354          * @param string|array $data
 355          * @return array|string Cleaned-up version of the given
 356          * @private
 357          */
 358         public function normalizeUnicode( $data ) {
 359                 if ( is_array( $data ) ) {
 360                         foreach ( $data as $key => $val ) {
 361                                 $data[$key] = $this->normalizeUnicode( $val );
 362                         }
 363                 } else {
 364                         global $wgContLang;
 365                         $data = isset( $wgContLang ) ?
 366                                 $wgContLang->normalize( $data ) :
 367                                 UtfNormal\Validator::cleanUp( $data );
 368                 }
 369                 return $data;
 370         }
 371
 372         /**
 373          * Fetch a value from the given array or return $default if it's not set.
 374          *
 375          * @param array $arr
 376          * @param string $name
 377          * @param mixed $default
 378          * @return mixed
 379          */
 380         private function getGPCVal( $arr, $name, $default ) {
 381                 # PHP is so nice to not touch input data, except sometimes:
 382                 # https://secure.php.net/variables.external#language.variables.external.dot-in-names
 383                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 384                 $name = strtr( $name, '.', '_' );
 385                 if ( isset( $arr[$name] ) ) {
 386                         global $wgContLang;
 387                         $data = $arr[$name];
 388                         if ( isset( $_GET[$name] ) && !is_array( $data ) ) {
 389                                 # Check for alternate/legacy character encoding.
 390                                 if ( isset( $wgContLang ) ) {
 391                                         $data = $wgContLang->checkTitleEncoding( $data );
 392                                 }
 393                         }
 394                         $data = $this->normalizeUnicode( $data );
 395                         return $data;
 396                 } else {
 397                         return $default;
 398                 }
 399         }
 400
 401         /**
 402          * Fetch a scalar from the input without normalization, or return $default
 403          * if it's not set.
 404          *
 405          * Unlike self::getVal(), this does not perform any normalization on the
 406          * input value.
 407          *
 408          * @since 1.28
 409          * @param string $name
 410          * @param string|null $default
 411          * @return string|null
 412          */
 413         public function getRawVal( $name, $default = null ) {
 414                 $name = strtr( $name, '.', '_' ); // See comment in self::getGPCVal()
 415                 if ( isset( $this->data[$name] ) && !is_array( $this->data[$name] ) ) {
 416                         $val = $this->data[$name];
 417                 } else {
 418                         $val = $default;
 419                 }
 420                 if ( is_null( $val ) ) {
 421                         return $val;
 422                 } else {
 423                         return (string)$val;
 424                 }
 425         }
 426
 427         /**
 428          * Fetch a scalar from the input or return $default if it's not set.
 429          * Returns a string. Arrays are discarded. Useful for
 430          * non-freeform text inputs (e.g. predefined internal text keys
 431          * selected by a drop-down menu). For freeform input, see getText().
 432          *
 433          * @param string $name
 434          * @param string $default Optional default (or null)
 435          * @return string|null
 436          */
 437         public function getVal( $name, $default = null ) {
 438                 $val = $this->getGPCVal( $this->data, $name, $default );
 439                 if ( is_array( $val ) ) {
 440                         $val = $default;
 441                 }
 442                 if ( is_null( $val ) ) {
 443                         return $val;
 444                 } else {
 445                         return (string)$val;
 446                 }
 447         }
 448
 449         /**
 450          * Set an arbitrary value into our get/post data.
 451          *
 452          * @param string $key Key name to use
 453          * @param mixed $value Value to set
 454          * @return mixed Old value if one was present, null otherwise
 455          */
 456         public function setVal( $key, $value ) {
 457                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 458                 $this->data[$key] = $value;
 459                 return $ret;
 460         }
 461
 462         /**
 463          * Unset an arbitrary value from our get/post data.
 464          *
 465          * @param string $key Key name to use
 466          * @return mixed Old value if one was present, null otherwise
 467          */
 468         public function unsetVal( $key ) {
 469                 if ( !isset( $this->data[$key] ) ) {
 470                         $ret = null;
 471                 } else {
 472                         $ret = $this->data[$key];
 473                         unset( $this->data[$key] );
 474                 }
 475                 return $ret;
 476         }
 477
 478         /**
 479          * Fetch an array from the input or return $default if it's not set.
 480          * If source was scalar, will return an array with a single element.
 481          * If no source and no default, returns null.
 482          *
 483          * @param string $name
 484          * @param array $default Optional default (or null)
 485          * @return array|null
 486          */
 487         public function getArray( $name, $default = null ) {
 488                 $val = $this->getGPCVal( $this->data, $name, $default );
 489                 if ( is_null( $val ) ) {
 490                         return null;
 491                 } else {
 492                         return (array)$val;
 493                 }
 494         }
 495
 496         /**
 497          * Fetch an array of integers, or return $default if it's not set.
 498          * If source was scalar, will return an array with a single element.
 499          * If no source and no default, returns null.
 500          * If an array is returned, contents are guaranteed to be integers.
 501          *
 502          * @param string $name
 503          * @param array $default Option default (or null)
 504          * @return array Array of ints
 505          */
 506         public function getIntArray( $name, $default = null ) {
 507                 $val = $this->getArray( $name, $default );
 508                 if ( is_array( $val ) ) {
 509                         $val = array_map( 'intval', $val );
 510                 }
 511                 return $val;
 512         }
 513
 514         /**
 515          * Fetch an integer value from the input or return $default if not set.
 516          * Guaranteed to return an integer; non-numeric input will typically
 517          * return 0.
 518          *
 519          * @param string $name
 520          * @param int $default
 521          * @return int
 522          */
 523         public function getInt( $name, $default = 0 ) {
 524                 return intval( $this->getRawVal( $name, $default ) );
 525         }
 526
 527         /**
 528          * Fetch an integer value from the input or return null if empty.
 529          * Guaranteed to return an integer or null; non-numeric input will
 530          * typically return null.
 531          *
 532          * @param string $name
 533          * @return int|null
 534          */
 535         public function getIntOrNull( $name ) {
 536                 $val = $this->getRawVal( $name );
 537                 return is_numeric( $val )
 538                         ? intval( $val )
 539                         : null;
 540         }
 541
 542         /**
 543          * Fetch a floating point value from the input or return $default if not set.
 544          * Guaranteed to return a float; non-numeric input will typically
 545          * return 0.
 546          *
 547          * @since 1.23
 548          * @param string $name
 549          * @param float $default
 550          * @return float
 551          */
 552         public function getFloat( $name, $default = 0.0 ) {
 553                 return floatval( $this->getRawVal( $name, $default ) );
 554         }
 555
 556         /**
 557          * Fetch a boolean value from the input or return $default if not set.
 558          * Guaranteed to return true or false, with normal PHP semantics for
 559          * boolean interpretation of strings.
 560          *
 561          * @param string $name
 562          * @param bool $default
 563          * @return bool
 564          */
 565         public function getBool( $name, $default = false ) {
 566                 return (bool)$this->getRawVal( $name, $default );
 567         }
 568
 569         /**
 570          * Fetch a boolean value from the input or return $default if not set.
 571          * Unlike getBool, the string "false" will result in boolean false, which is
 572          * useful when interpreting information sent from JavaScript.
 573          *
 574          * @param string $name
 575          * @param bool $default
 576          * @return bool
 577          */
 578         public function getFuzzyBool( $name, $default = false ) {
 579                 return $this->getBool( $name, $default )
 580                         && strcasecmp( $this->getRawVal( $name ), 'false' ) !== 0;
 581         }
 582
 583         /**
 584          * Return true if the named value is set in the input, whatever that
 585          * value is (even "0"). Return false if the named value is not set.
 586          * Example use is checking for the presence of check boxes in forms.
 587          *
 588          * @param string $name
 589          * @return bool
 590          */
 591         public function getCheck( $name ) {
 592                 # Checkboxes and buttons are only present when clicked
 593                 # Presence connotes truth, absence false
 594                 return $this->getRawVal( $name, null ) !== null;
 595         }
 596
 597         /**
 598          * Fetch a text string from the given array or return $default if it's not
 599          * set. Carriage returns are stripped from the text. This should generally
 600          * be used for form "<textarea>" and "<input>" fields, and for
 601          * user-supplied freeform text input.
 602          *
 603          * @param string $name
 604          * @param string $default Optional
 605          * @return string
 606          */
 607         public function getText( $name, $default = '' ) {
 608                 $val = $this->getVal( $name, $default );
 609                 return str_replace( "\r\n", "\n", $val );
 610         }
 611
 612         /**
 613          * Extracts the given named values into an array.
 614          * If no arguments are given, returns all input values.
 615          * No transformation is performed on the values.
 616          *
 617          * @return array
 618          */
 619         public function getValues() {
 620                 $names = func_get_args();
 621                 if ( count( $names ) == 0 ) {
 622                         $names = array_keys( $this->data );
 623                 }
 624
 625                 $retVal = [];
 626                 foreach ( $names as $name ) {
 627                         $value = $this->getGPCVal( $this->data, $name, null );
 628                         if ( !is_null( $value ) ) {
 629                                 $retVal[$name] = $value;
 630                         }
 631                 }
 632                 return $retVal;
 633         }
 634
 635         /**
 636          * Returns the names of all input values excluding those in $exclude.
 637          *
 638          * @param array $exclude
 639          * @return array
 640          */
 641         public function getValueNames( $exclude = [] ) {
 642                 return array_diff( array_keys( $this->getValues() ), $exclude );
 643         }
 644
 645         /**
 646          * Get the values passed in the query string.
 647          * No transformation is performed on the values.
 648          *
 649          * @codeCoverageIgnore
 650          * @return array
 651          */
 652         public function getQueryValues() {
 653                 return $_GET;
 654         }
 655
 656         /**
 657          * Return the contents of the Query with no decoding. Use when you need to
 658          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 659          *
 660          * @codeCoverageIgnore
 661          * @return string
 662          */
 663         public function getRawQueryString() {
 664                 return $_SERVER['QUERY_STRING'];
 665         }
 666
 667         /**
 668          * Return the contents of the POST with no decoding. Use when you need to
 669          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 670          *
 671          * @return string
 672          */
 673         public function getRawPostString() {
 674                 if ( !$this->wasPosted() ) {
 675                         return '';
 676                 }
 677                 return $this->getRawInput();
 678         }
 679
 680         /**
 681          * Return the raw request body, with no processing. Cached since some methods
 682          * disallow reading the stream more than once. As stated in the php docs, this
 683          * does not work with enctype="multipart/form-data".
 684          *
 685          * @return string
 686          */
 687         public function getRawInput() {
 688                 static $input = null;
 689                 if ( $input === null ) {
 690                         $input = file_get_contents( 'php://input' );
 691                 }
 692                 return $input;
 693         }
 694
 695         /**
 696          * Get the HTTP method used for this request.
 697          *
 698          * @return string
 699          */
 700         public function getMethod() {
 701                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 702         }
 703
 704         /**
 705          * Returns true if the present request was reached by a POST operation,
 706          * false otherwise (GET, HEAD, or command-line).
 707          *
 708          * Note that values retrieved by the object may come from the
 709          * GET URL etc even on a POST request.
 710          *
 711          * @return bool
 712          */
 713         public function wasPosted() {
 714                 return $this->getMethod() == 'POST';
 715         }
 716
 717         /**
 718          * Return the session for this request
 719          *
 720          * This might unpersist an existing session if it was invalid.
 721          *
 722          * @since 1.27
 723          * @note For performance, keep the session locally if you will be making
 724          *  much use of it instead of calling this method repeatedly.
 725          * @return Session
 726          */
 727         public function getSession() {
 728                 if ( $this->sessionId !== null ) {
 729                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 730                         if ( $session ) {
 731                                 return $session;
 732                         }
 733                 }
 734
 735                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 736                 $this->sessionId = $session->getSessionId();
 737                 return $session;
 738         }
 739
 740         /**
 741          * Set the session for this request
 742          * @since 1.27
 743          * @private For use by MediaWiki\Session classes only
 744          * @param SessionId $sessionId
 745          */
 746         public function setSessionId( SessionId $sessionId ) {
 747                 $this->sessionId = $sessionId;
 748         }
 749
 750         /**
 751          * Get the session id for this request, if any
 752          * @since 1.27
 753          * @private For use by MediaWiki\Session classes only
 754          * @return SessionId|null
 755          */
 756         public function getSessionId() {
 757                 return $this->sessionId;
 758         }
 759
 760         /**
 761          * Get a cookie from the $_COOKIE jar
 762          *
 763          * @param string $key The name of the cookie
 764          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 765          * @param mixed $default What to return if the value isn't found
 766          * @return mixed Cookie value or $default if the cookie not set
 767          */
 768         public function getCookie( $key, $prefix = null, $default = null ) {
 769                 if ( $prefix === null ) {
 770                         global $wgCookiePrefix;
 771                         $prefix = $wgCookiePrefix;
 772                 }
 773                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 774         }
 775
 776         /**
 777          * Return the path and query string portion of the main request URI.
 778          * This will be suitable for use as a relative link in HTML output.
 779          *
 780          * @throws MWException
 781          * @return string
 782          */
 783         public static function getGlobalRequestURL() {
 784                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 785                         $base = $_SERVER['REQUEST_URI'];
 786                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 787                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 788                 ) {
 789                         // Probably IIS; doesn't set REQUEST_URI
 790                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 791                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 792                         $base = $_SERVER['SCRIPT_NAME'];
 793                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 794                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 795                         }
 796                 } else {
 797                         // This shouldn't happen!
 798                         throw new MWException( "Web server doesn't provide either " .
 799                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 800                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 801                 }
 802                 // User-agents should not send a fragment with the URI, but
 803                 // if they do, and the web server passes it on to us, we
 804                 // need to strip it or we get false-positive redirect loops
 805                 // or weird output URLs
 806                 $hash = strpos( $base, '#' );
 807                 if ( $hash !== false ) {
 808                         $base = substr( $base, 0, $hash );
 809                 }
 810
 811                 if ( $base[0] == '/' ) {
 812                         // More than one slash will look like it is protocol relative
 813                         return preg_replace( '!^/+!', '/', $base );
 814                 } else {
 815                         // We may get paths with a host prepended; strip it.
 816                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 817                 }
 818         }
 819
 820         /**
 821          * Return the path and query string portion of the request URI.
 822          * This will be suitable for use as a relative link in HTML output.
 823          *
 824          * @throws MWException
 825          * @return string
 826          */
 827         public function getRequestURL() {
 828                 return self::getGlobalRequestURL();
 829         }
 830
 831         /**
 832          * Return the request URI with the canonical service and hostname, path,
 833          * and query string. This will be suitable for use as an absolute link
 834          * in HTML or other output.
 835          *
 836          * If $wgServer is protocol-relative, this will return a fully
 837          * qualified URL with the protocol that was used for this request.
 838          *
 839          * @return string
 840          */
 841         public function getFullRequestURL() {
 842                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 843         }
 844
 845         /**
 846          * @param string $key
 847          * @param string $value
 848          * @return string
 849          */
 850         public function appendQueryValue( $key, $value ) {
 851                 return $this->appendQueryArray( [ $key => $value ] );
 852         }
 853
 854         /**
 855          * Appends or replaces value of query variables.
 856          *
 857          * @param array $array Array of values to replace/add to query
 858          * @return string
 859          */
 860         public function appendQueryArray( $array ) {
 861                 $newquery = $this->getQueryValues();
 862                 unset( $newquery['title'] );
 863                 $newquery = array_merge( $newquery, $array );
 864
 865                 return wfArrayToCgi( $newquery );
 866         }
 867
 868         /**
 869          * Check for limit and offset parameters on the input, and return sensible
 870          * defaults if not given. The limit must be positive and is capped at 5000.
 871          * Offset must be positive but is not capped.
 872          *
 873          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 874          * @param string $optionname To specify an option other than rclimit to pull from.
 875          * @return int[] First element is limit, second is offset
 876          */
 877         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 878                 global $wgUser;
 879
 880                 $limit = $this->getInt( 'limit', 0 );
 881                 if ( $limit < 0 ) {
 882                         $limit = 0;
 883                 }
 884                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 885                         $limit = $wgUser->getIntOption( $optionname );
 886                 }
 887                 if ( $limit <= 0 ) {
 888                         $limit = $deflimit;
 889                 }
 890                 if ( $limit > 5000 ) {
 891                         $limit = 5000; # We have *some* limits...
 892                 }
 893
 894                 $offset = $this->getInt( 'offset', 0 );
 895                 if ( $offset < 0 ) {
 896                         $offset = 0;
 897                 }
 898
 899                 return [ $limit, $offset ];
 900         }
 901
 902         /**
 903          * Return the path to the temporary file where PHP has stored the upload.
 904          *
 905          * @param string $key
 906          * @return string|null String or null if no such file.
 907          */
 908         public function getFileTempname( $key ) {
 909                 $file = new WebRequestUpload( $this, $key );
 910                 return $file->getTempName();
 911         }
 912
 913         /**
 914          * Return the upload error or 0
 915          *
 916          * @param string $key
 917          * @return int
 918          */
 919         public function getUploadError( $key ) {
 920                 $file = new WebRequestUpload( $this, $key );
 921                 return $file->getError();
 922         }
 923
 924         /**
 925          * Return the original filename of the uploaded file, as reported by
 926          * the submitting user agent. HTML-style character entities are
 927          * interpreted and normalized to Unicode normalization form C, in part
 928          * to deal with weird input from Safari with non-ASCII filenames.
 929          *
 930          * Other than this the name is not verified for being a safe filename.
 931          *
 932          * @param string $key
 933          * @return string|null String or null if no such file.
 934          */
 935         public function getFileName( $key ) {
 936                 $file = new WebRequestUpload( $this, $key );
 937                 return $file->getName();
 938         }
 939
 940         /**
 941          * Return a WebRequestUpload object corresponding to the key
 942          *
 943          * @param string $key
 944          * @return WebRequestUpload
 945          */
 946         public function getUpload( $key ) {
 947                 return new WebRequestUpload( $this, $key );
 948         }
 949
 950         /**
 951          * Return a handle to WebResponse style object, for setting cookies,
 952          * headers and other stuff, for Request being worked on.
 953          *
 954          * @return WebResponse
 955          */
 956         public function response() {
 957                 /* Lazy initialization of response object for this request */
 958                 if ( !is_object( $this->response ) ) {
 959                         $class = ( $this instanceof FauxRequest ) ? FauxResponse::class : WebResponse::class;
 960                         $this->response = new $class();
 961                 }
 962                 return $this->response;
 963         }
 964
 965         /**
 966          * Initialise the header list
 967          */
 968         protected function initHeaders() {
 969                 if ( count( $this->headers ) ) {
 970                         return;
 971                 }
 972
 973                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 974                 if ( $apacheHeaders ) {
 975                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 976                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 977                         }
 978                 } else {
 979                         foreach ( $_SERVER as $name => $value ) {
 980                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 981                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 982                                         $this->headers[$name] = $value;
 983                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 984                                         $this->headers['CONTENT-LENGTH'] = $value;
 985                                 }
 986                         }
 987                 }
 988         }
 989
 990         /**
 991          * Get an array containing all request headers
 992          *
 993          * @return array Mapping header name to its value
 994          */
 995         public function getAllHeaders() {
 996                 $this->initHeaders();
 997                 return $this->headers;
 998         }
 999
1000         /**
1001          * Get a request header, or false if it isn't set.
1002          *
1003          * @param string $name Case-insensitive header name
1004          * @param int $flags Bitwise combination of:
1005          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
1006          *                               of values, as described in RFC 2616 § 4.2.
1007          *                               (since 1.26).
1008          * @return string|array|bool False if header is unset; otherwise the
1009          *  header value(s) as either a string (the default) or an array, if
1010          *  WebRequest::GETHEADER_LIST flag was set.
1011          */
1012         public function getHeader( $name, $flags = 0 ) {
1013                 $this->initHeaders();
1014                 $name = strtoupper( $name );
1015                 if ( !isset( $this->headers[$name] ) ) {
1016                         return false;
1017                 }
1018                 $value = $this->headers[$name];
1019                 if ( $flags & self::GETHEADER_LIST ) {
1020                         $value = array_map( 'trim', explode( ',', $value ) );
1021                 }
1022                 return $value;
1023         }
1024
1025         /**
1026          * Get data from the session
1027          *
1028          * @note Prefer $this->getSession() instead if making multiple calls.
1029          * @param string $key Name of key in the session
1030          * @return mixed
1031          */
1032         public function getSessionData( $key ) {
1033                 return $this->getSession()->get( $key );
1034         }
1035
1036         /**
1037          * Set session data
1038          *
1039          * @note Prefer $this->getSession() instead if making multiple calls.
1040          * @param string $key Name of key in the session
1041          * @param mixed $data
1042          */
1043         public function setSessionData( $key, $data ) {
1044                 $this->getSession()->set( $key, $data );
1045         }
1046
1047         /**
1048          * Check if Internet Explorer will detect an incorrect cache extension in
1049          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1050          * message or redirect to a safer URL. Returns true if the URL is OK, and
1051          * false if an error message has been shown and the request should be aborted.
1052          *
1053          * @param array $extWhitelist
1054          * @throws HttpError
1055          * @return bool
1056          */
1057         public function checkUrlExtension( $extWhitelist = [] ) {
1058                 $extWhitelist[] = 'php';
1059                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1060                         if ( !$this->wasPosted() ) {
1061                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1062                                         $this->getFullRequestURL(), $extWhitelist );
1063                                 if ( $newUrl !== false ) {
1064                                         $this->doSecurityRedirect( $newUrl );
1065                                         return false;
1066                                 }
1067                         }
1068                         throw new HttpError( 403,
1069                                 'Invalid file extension found in the path info or query string.' );
1070                 }
1071                 return true;
1072         }
1073
1074         /**
1075          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1076          * IE 6. Returns true if it was successful, false otherwise.
1077          *
1078          * @param string $url
1079          * @return bool
1080          */
1081         protected function doSecurityRedirect( $url ) {
1082                 header( 'Location: ' . $url );
1083                 header( 'Content-Type: text/html' );
1084                 $encUrl = htmlspecialchars( $url );
1085                 echo <<<HTML
1086 <!DOCTYPE html>
1087 <html>
1088 <head>
1089 <title>Security redirect</title>
1090 </head>
1091 <body>
1092 <h1>Security redirect</h1>
1093 <p>
1094 We can't serve non-HTML content from the URL you have requested, because
1095 Internet Explorer would interpret it as an incorrect and potentially dangerous
1096 content type.</p>
1097 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1098 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1099 Explorer from seeing a bogus file extension.
1100 </p>
1101 </body>
1102 </html>
1103 HTML;
1104                 echo "\n";
1105                 return true;
1106         }
1107
1108         /**
1109          * Parse the Accept-Language header sent by the client into an array
1110          *
1111          * @return array Array( languageCode => q-value ) sorted by q-value in
1112          *   descending order then appearing time in the header in ascending order.
1113          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1114          * This is aligned with rfc2616 section 14.4
1115          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1116          */
1117         public function getAcceptLang() {
1118                 // Modified version of code found at
1119                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1120                 $acceptLang = $this->getHeader( 'Accept-Language' );
1121                 if ( !$acceptLang ) {
1122                         return [];
1123                 }
1124
1125                 // Return the language codes in lower case
1126                 $acceptLang = strtolower( $acceptLang );
1127
1128                 // Break up string into pieces (languages and q factors)
1129                 $lang_parse = null;
1130                 preg_match_all(
1131                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1132                         $acceptLang,
1133                         $lang_parse
1134                 );
1135
1136                 if ( !count( $lang_parse[1] ) ) {
1137                         return [];
1138                 }
1139
1140                 $langcodes = $lang_parse[1];
1141                 $qvalues = $lang_parse[4];
1142                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1143
1144                 // Set default q factor to 1
1145                 foreach ( $indices as $index ) {
1146                         if ( $qvalues[$index] === '' ) {
1147                                 $qvalues[$index] = 1;
1148                         } elseif ( $qvalues[$index] == 0 ) {
1149                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1150                         }
1151                 }
1152
1153                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1154                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1155
1156                 // Create a list like "en" => 0.8
1157                 $langs = array_combine( $langcodes, $qvalues );
1158
1159                 return $langs;
1160         }
1161
1162         /**
1163          * Fetch the raw IP from the request
1164          *
1165          * @since 1.19
1166          *
1167          * @throws MWException
1168          * @return string
1169          */
1170         protected function getRawIP() {
1171                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1172                         return null;
1173                 }
1174
1175                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1176                         throw new MWException( __METHOD__
1177                                 . " : Could not determine the remote IP address due to multiple values." );
1178                 } else {
1179                         $ipchain = $_SERVER['REMOTE_ADDR'];
1180                 }
1181
1182                 return IP::canonicalize( $ipchain );
1183         }
1184
1185         /**
1186          * Work out the IP address based on various globals
1187          * For trusted proxies, use the XFF client IP (first of the chain)
1188          *
1189          * @since 1.19
1190          *
1191          * @throws MWException
1192          * @return string
1193          */
1194         public function getIP() {
1195                 global $wgUsePrivateIPs;
1196
1197                 # Return cached result
1198                 if ( $this->ip !== null ) {
1199                         return $this->ip;
1200                 }
1201
1202                 # collect the originating ips
1203                 $ip = $this->getRawIP();
1204                 if ( !$ip ) {
1205                         throw new MWException( 'Unable to determine IP.' );
1206                 }
1207
1208                 # Append XFF
1209                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1210                 if ( $forwardedFor !== false ) {
1211                         $proxyLookup = MediaWikiServices::getInstance()->getProxyLookup();
1212                         $isConfigured = $proxyLookup->isConfiguredProxy( $ip );
1213                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1214                         $ipchain = array_reverse( $ipchain );
1215                         array_unshift( $ipchain, $ip );
1216
1217                         # Step through XFF list and find the last address in the list which is a
1218                         # trusted server. Set $ip to the IP address given by that trusted server,
1219                         # unless the address is not sensible (e.g. private). However, prefer private
1220                         # IP addresses over proxy servers controlled by this site (more sensible).
1221                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1222                         foreach ( $ipchain as $i => $curIP ) {
1223                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1224                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1225                                         || !$proxyLookup->isTrustedProxy( $curIP )
1226                                 ) {
1227                                         break; // IP is not valid/trusted or does not point to anything
1228                                 }
1229                                 if (
1230                                         IP::isPublic( $ipchain[$i + 1] ) ||
1231                                         $wgUsePrivateIPs ||
1232                                         $proxyLookup->isConfiguredProxy( $curIP ) // T50919; treat IP as sane
1233                                 ) {
1234                                         // Follow the next IP according to the proxy
1235                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1236                                         if ( !$nextIP && $isConfigured ) {
1237                                                 // We have not yet made it past CDN/proxy servers of this site,
1238                                                 // so either they are misconfigured or there is some IP spoofing.
1239                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1240                                         }
1241                                         $ip = $nextIP;
1242                                         // keep traversing the chain
1243                                         continue;
1244                                 }
1245                                 break;
1246                         }
1247                 }
1248
1249                 # Allow extensions to improve our guess
1250                 Hooks::run( 'GetIP', [ &$ip ] );
1251
1252                 if ( !$ip ) {
1253                         throw new MWException( "Unable to determine IP." );
1254                 }
1255
1256                 wfDebug( "IP: $ip\n" );
1257                 $this->ip = $ip;
1258                 return $ip;
1259         }
1260
1261         /**
1262          * @param string $ip
1263          * @return void
1264          * @since 1.21
1265          */
1266         public function setIP( $ip ) {
1267                 $this->ip = $ip;
1268         }
1269
1270         /**
1271          * Check if this request uses a "safe" HTTP method
1272          *
1273          * Safe methods are verbs (e.g. GET/HEAD/OPTIONS) used for obtaining content. Such requests
1274          * are not expected to mutate content, especially in ways attributable to the client. Verbs
1275          * like POST and PUT are typical of non-safe requests which often change content.
1276          *
1277          * @return bool
1278          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1279          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1280          * @since 1.28
1281          */
1282         public function hasSafeMethod() {
1283                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1284                         return false; // CLI mode
1285                 }
1286
1287                 return in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS', 'TRACE' ] );
1288         }
1289
1290         /**
1291          * Whether this request should be identified as being "safe"
1292          *
1293          * This means that the client is not requesting any state changes and that database writes
1294          * are not inherently required. Ideally, no visible updates would happen at all. If they
1295          * must, then they should not be publically attributed to the end user.
1296          *
1297          * In more detail:
1298          *   - Cache populations and refreshes MAY occur.
1299          *   - Private user session updates and private server logging MAY occur.
1300          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1301          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1302          *
1303          * @return bool
1304          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1305          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1306          * @since 1.28
1307          */
1308         public function isSafeRequest() {
1309                 if ( $this->markedAsSafe && $this->wasPosted() ) {
1310                         return true; // marked as a "safe" POST
1311                 }
1312
1313                 return $this->hasSafeMethod();
1314         }
1315
1316         /**
1317          * Mark this request as identified as being nullipotent even if it is a POST request
1318          *
1319          * POST requests are often used due to the need for a client payload, even if the request
1320          * is otherwise equivalent to a "safe method" request.
1321          *
1322          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1323          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1324          * @since 1.28
1325          */
1326         public function markAsSafeRequest() {
1327                 $this->markedAsSafe = true;
1328         }
1329 }