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