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\Session\SessionManager;
  27
  28 /**
  29  * The WebRequest class encapsulates getting at data passed in the
  30  * URL or via a POSTed form stripping illegal input characters and
  31  * normalizing Unicode sequences.
  32  *
  33  * @ingroup HTTP
  34  */
  35 class WebRequest {
  36         protected $data, $headers = array();
  37
  38         /**
  39          * Flag to make WebRequest::getHeader return an array of values.
  40          * @since 1.26
  41          */
  42         const GETHEADER_LIST = 1;
  43
  44         /**
  45          * Lazy-init response object
  46          * @var WebResponse
  47          */
  48         private $response;
  49
  50         /**
  51          * Cached client IP address
  52          * @var string
  53          */
  54         private $ip;
  55
  56         /**
  57          * The timestamp of the start of the request, with microsecond precision.
  58          * @var float
  59          */
  60         protected $requestTime;
  61
  62         /**
  63          * Cached URL protocol
  64          * @var string
  65          */
  66         protected $protocol;
  67
  68         /**
  69          * @var \\MediaWiki\\Session\\SessionId|null Session ID to use for this
  70          *  request. We can't save the session directly due to reference cycles not
  71          *  working too well (slow GC in Zend and never collected in HHVM).
  72          */
  73         protected $sessionId = null;
  74
  75         public function __construct() {
  76                 $this->requestTime = isset( $_SERVER['REQUEST_TIME_FLOAT'] )
  77                         ? $_SERVER['REQUEST_TIME_FLOAT'] : microtime( true );
  78
  79                 // POST overrides GET data
  80                 // We don't use $_REQUEST here to avoid interference from cookies...
  81                 $this->data = $_POST + $_GET;
  82         }
  83
  84         /**
  85          * Extract relevant query arguments from the http request uri's path
  86          * to be merged with the normal php provided query arguments.
  87          * Tries to use the REQUEST_URI data if available and parses it
  88          * according to the wiki's configuration looking for any known pattern.
  89          *
  90          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
  91          * provided by the server if any and use that to set a 'title' parameter.
  92          *
  93          * @param string $want If this is not 'all', then the function
  94          * will return an empty array if it determines that the URL is
  95          * inside a rewrite path.
  96          *
  97          * @return array Any query arguments found in path matches.
  98          */
  99         public static function getPathInfo( $want = 'all' ) {
 100                 global $wgUsePathInfo;
 101                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
 102                 // And also by Apache 2.x, double slashes are converted to single slashes.
 103                 // So we will use REQUEST_URI if possible.
 104                 $matches = array();
 105                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
 106                         // Slurp out the path portion to examine...
 107                         $url = $_SERVER['REQUEST_URI'];
 108                         if ( !preg_match( '!^https?://!', $url ) ) {
 109                                 $url = 'http://unused' . $url;
 110                         }
 111                         MediaWiki\suppressWarnings();
 112                         $a = parse_url( $url );
 113                         MediaWiki\restoreWarnings();
 114                         if ( $a ) {
 115                                 $path = isset( $a['path'] ) ? $a['path'] : '';
 116
 117                                 global $wgScript;
 118                                 if ( $path == $wgScript && $want !== 'all' ) {
 119                                         // Script inside a rewrite path?
 120                                         // Abort to keep from breaking...
 121                                         return $matches;
 122                                 }
 123
 124                                 $router = new PathRouter;
 125
 126                                 // Raw PATH_INFO style
 127                                 $router->add( "$wgScript/$1" );
 128
 129                                 if ( isset( $_SERVER['SCRIPT_NAME'] )
 130                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] )
 131                                 ) {
 132                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 133                                         # But we do have some other .php files such as img_auth.php
 134                                         # Don't let root article paths clober the parsing for them
 135                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 136                                 }
 137
 138                                 global $wgArticlePath;
 139                                 if ( $wgArticlePath ) {
 140                                         $router->add( $wgArticlePath );
 141                                 }
 142
 143                                 global $wgActionPaths;
 144                                 if ( $wgActionPaths ) {
 145                                         $router->add( $wgActionPaths, array( 'action' => '$key' ) );
 146                                 }
 147
 148                                 global $wgVariantArticlePath, $wgContLang;
 149                                 if ( $wgVariantArticlePath ) {
 150                                         $router->add( $wgVariantArticlePath,
 151                                                 array( 'variant' => '$2' ),
 152                                                 array( '$2' => $wgContLang->getVariants() )
 153                                         );
 154                                 }
 155
 156                                 Hooks::run( 'WebRequestPathInfoRouter', array( $router ) );
 157
 158                                 $matches = $router->parse( $path );
 159                         }
 160                 } elseif ( $wgUsePathInfo ) {
 161                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 162                                 // Mangled PATH_INFO
 163                                 // http://bugs.php.net/bug.php?id=31892
 164                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 165                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 166
 167                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && $_SERVER['PATH_INFO'] != '' ) {
 168                                 // Regular old PATH_INFO yay
 169                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 170                         }
 171                 }
 172
 173                 return $matches;
 174         }
 175
 176         /**
 177          * Work out an appropriate URL prefix containing scheme and host, based on
 178          * information detected from $_SERVER
 179          *
 180          * @return string
 181          */
 182         public static function detectServer() {
 183                 global $wgAssumeProxiesUseDefaultProtocolPorts;
 184
 185                 $proto = self::detectProtocol();
 186                 $stdPort = $proto === 'https' ? 443 : 80;
 187
 188                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 189                 $host = 'localhost';
 190                 $port = $stdPort;
 191                 foreach ( $varNames as $varName ) {
 192                         if ( !isset( $_SERVER[$varName] ) ) {
 193                                 continue;
 194                         }
 195
 196                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 197                         if ( !$parts ) {
 198                                 // Invalid, do not use
 199                                 continue;
 200                         }
 201
 202                         $host = $parts[0];
 203                         if ( $wgAssumeProxiesUseDefaultProtocolPorts && isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) ) {
 204                                 // Bug 70021: Assume that upstream proxy is running on the default
 205                                 // port based on the protocol. We have no reliable way to determine
 206                                 // the actual port in use upstream.
 207                                 $port = $stdPort;
 208                         } elseif ( $parts[1] === false ) {
 209                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 210                                         $port = $_SERVER['SERVER_PORT'];
 211                                 } // else leave it as $stdPort
 212                         } else {
 213                                 $port = $parts[1];
 214                         }
 215                         break;
 216                 }
 217
 218                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 219         }
 220
 221         /**
 222          * Detect the protocol from $_SERVER.
 223          * This is for use prior to Setup.php, when no WebRequest object is available.
 224          * At other times, use the non-static function getProtocol().
 225          *
 226          * @return array
 227          */
 228         public static function detectProtocol() {
 229                 if ( ( !empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) ||
 230                         ( isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) &&
 231                         $_SERVER['HTTP_X_FORWARDED_PROTO'] === 'https' ) ) {
 232                         return 'https';
 233                 } else {
 234                         return 'http';
 235                 }
 236         }
 237
 238         /**
 239          * Get the number of seconds to have elapsed since request start,
 240          * in fractional seconds, with microsecond resolution.
 241          *
 242          * @return float
 243          * @since 1.25
 244          */
 245         public function getElapsedTime() {
 246                 return microtime( true ) - $this->requestTime;
 247         }
 248
 249         /**
 250          * Get the current URL protocol (http or https)
 251          * @return string
 252          */
 253         public function getProtocol() {
 254                 if ( $this->protocol === null ) {
 255                         $this->protocol = self::detectProtocol();
 256                 }
 257                 return $this->protocol;
 258         }
 259
 260         /**
 261          * Check for title, action, and/or variant data in the URL
 262          * and interpolate it into the GET variables.
 263          * This should only be run after $wgContLang is available,
 264          * as we may need the list of language variants to determine
 265          * available variant URLs.
 266          */
 267         public function interpolateTitle() {
 268                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 269                 if ( defined( 'MW_API' ) ) {
 270                         return;
 271                 }
 272
 273                 $matches = self::getPathInfo( 'title' );
 274                 foreach ( $matches as $key => $val ) {
 275                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 276                 }
 277         }
 278
 279         /**
 280          * URL rewriting function; tries to extract page title and,
 281          * optionally, one other fixed parameter value from a URL path.
 282          *
 283          * @param string $path The URL path given from the client
 284          * @param array $bases One or more URLs, optionally with $1 at the end
 285          * @param string $key If provided, the matching key in $bases will be
 286          *    passed on as the value of this URL parameter
 287          * @return array Array of URL variables to interpolate; empty if no match
 288          */
 289         static function extractTitle( $path, $bases, $key = false ) {
 290                 foreach ( (array)$bases as $keyValue => $base ) {
 291                         // Find the part after $wgArticlePath
 292                         $base = str_replace( '$1', '', $base );
 293                         $baseLen = strlen( $base );
 294                         if ( substr( $path, 0, $baseLen ) == $base ) {
 295                                 $raw = substr( $path, $baseLen );
 296                                 if ( $raw !== '' ) {
 297                                         $matches = array( 'title' => rawurldecode( $raw ) );
 298                                         if ( $key ) {
 299                                                 $matches[$key] = $keyValue;
 300                                         }
 301                                         return $matches;
 302                                 }
 303                         }
 304                 }
 305                 return array();
 306         }
 307
 308         /**
 309          * Recursively normalizes UTF-8 strings in the given array.
 310          *
 311          * @param string|array $data
 312          * @return array|string Cleaned-up version of the given
 313          * @private
 314          */
 315         function normalizeUnicode( $data ) {
 316                 if ( is_array( $data ) ) {
 317                         foreach ( $data as $key => $val ) {
 318                                 $data[$key] = $this->normalizeUnicode( $val );
 319                         }
 320                 } else {
 321                         global $wgContLang;
 322                         $data = isset( $wgContLang ) ?
 323                                 $wgContLang->normalize( $data ) :
 324                                 UtfNormal\Validator::cleanUp( $data );
 325                 }
 326                 return $data;
 327         }
 328
 329         /**
 330          * Fetch a value from the given array or return $default if it's not set.
 331          *
 332          * @param array $arr
 333          * @param string $name
 334          * @param mixed $default
 335          * @return mixed
 336          */
 337         private function getGPCVal( $arr, $name, $default ) {
 338                 # PHP is so nice to not touch input data, except sometimes:
 339                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 340                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 341                 $name = strtr( $name, '.', '_' );
 342                 if ( isset( $arr[$name] ) ) {
 343                         global $wgContLang;
 344                         $data = $arr[$name];
 345                         if ( isset( $_GET[$name] ) && !is_array( $data ) ) {
 346                                 # Check for alternate/legacy character encoding.
 347                                 if ( isset( $wgContLang ) ) {
 348                                         $data = $wgContLang->checkTitleEncoding( $data );
 349                                 }
 350                         }
 351                         $data = $this->normalizeUnicode( $data );
 352                         return $data;
 353                 } else {
 354                         return $default;
 355                 }
 356         }
 357
 358         /**
 359          * Fetch a scalar from the input or return $default if it's not set.
 360          * Returns a string. Arrays are discarded. Useful for
 361          * non-freeform text inputs (e.g. predefined internal text keys
 362          * selected by a drop-down menu). For freeform input, see getText().
 363          *
 364          * @param string $name
 365          * @param string $default Optional default (or null)
 366          * @return string
 367          */
 368         public function getVal( $name, $default = null ) {
 369                 $val = $this->getGPCVal( $this->data, $name, $default );
 370                 if ( is_array( $val ) ) {
 371                         $val = $default;
 372                 }
 373                 if ( is_null( $val ) ) {
 374                         return $val;
 375                 } else {
 376                         return (string)$val;
 377                 }
 378         }
 379
 380         /**
 381          * Set an arbitrary value into our get/post data.
 382          *
 383          * @param string $key Key name to use
 384          * @param mixed $value Value to set
 385          * @return mixed Old value if one was present, null otherwise
 386          */
 387         public function setVal( $key, $value ) {
 388                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 389                 $this->data[$key] = $value;
 390                 return $ret;
 391         }
 392
 393         /**
 394          * Unset an arbitrary value from our get/post data.
 395          *
 396          * @param string $key Key name to use
 397          * @return mixed Old value if one was present, null otherwise
 398          */
 399         public function unsetVal( $key ) {
 400                 if ( !isset( $this->data[$key] ) ) {
 401                         $ret = null;
 402                 } else {
 403                         $ret = $this->data[$key];
 404                         unset( $this->data[$key] );
 405                 }
 406                 return $ret;
 407         }
 408
 409         /**
 410          * Fetch an array from the input or return $default if it's not set.
 411          * If source was scalar, will return an array with a single element.
 412          * If no source and no default, returns null.
 413          *
 414          * @param string $name
 415          * @param array $default Optional default (or null)
 416          * @return array
 417          */
 418         public function getArray( $name, $default = null ) {
 419                 $val = $this->getGPCVal( $this->data, $name, $default );
 420                 if ( is_null( $val ) ) {
 421                         return null;
 422                 } else {
 423                         return (array)$val;
 424                 }
 425         }
 426
 427         /**
 428          * Fetch an array of integers, or return $default if it's not set.
 429          * If source was scalar, will return an array with a single element.
 430          * If no source and no default, returns null.
 431          * If an array is returned, contents are guaranteed to be integers.
 432          *
 433          * @param string $name
 434          * @param array $default Option default (or null)
 435          * @return array Array of ints
 436          */
 437         public function getIntArray( $name, $default = null ) {
 438                 $val = $this->getArray( $name, $default );
 439                 if ( is_array( $val ) ) {
 440                         $val = array_map( 'intval', $val );
 441                 }
 442                 return $val;
 443         }
 444
 445         /**
 446          * Fetch an integer value from the input or return $default if not set.
 447          * Guaranteed to return an integer; non-numeric input will typically
 448          * return 0.
 449          *
 450          * @param string $name
 451          * @param int $default
 452          * @return int
 453          */
 454         public function getInt( $name, $default = 0 ) {
 455                 return intval( $this->getVal( $name, $default ) );
 456         }
 457
 458         /**
 459          * Fetch an integer value from the input or return null if empty.
 460          * Guaranteed to return an integer or null; non-numeric input will
 461          * typically return null.
 462          *
 463          * @param string $name
 464          * @return int|null
 465          */
 466         public function getIntOrNull( $name ) {
 467                 $val = $this->getVal( $name );
 468                 return is_numeric( $val )
 469                         ? intval( $val )
 470                         : null;
 471         }
 472
 473         /**
 474          * Fetch a floating point value from the input or return $default if not set.
 475          * Guaranteed to return a float; non-numeric input will typically
 476          * return 0.
 477          *
 478          * @since 1.23
 479          * @param string $name
 480          * @param float $default
 481          * @return float
 482          */
 483         public function getFloat( $name, $default = 0.0 ) {
 484                 return floatval( $this->getVal( $name, $default ) );
 485         }
 486
 487         /**
 488          * Fetch a boolean value from the input or return $default if not set.
 489          * Guaranteed to return true or false, with normal PHP semantics for
 490          * boolean interpretation of strings.
 491          *
 492          * @param string $name
 493          * @param bool $default
 494          * @return bool
 495          */
 496         public function getBool( $name, $default = false ) {
 497                 return (bool)$this->getVal( $name, $default );
 498         }
 499
 500         /**
 501          * Fetch a boolean value from the input or return $default if not set.
 502          * Unlike getBool, the string "false" will result in boolean false, which is
 503          * useful when interpreting information sent from JavaScript.
 504          *
 505          * @param string $name
 506          * @param bool $default
 507          * @return bool
 508          */
 509         public function getFuzzyBool( $name, $default = false ) {
 510                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 511         }
 512
 513         /**
 514          * Return true if the named value is set in the input, whatever that
 515          * value is (even "0"). Return false if the named value is not set.
 516          * Example use is checking for the presence of check boxes in forms.
 517          *
 518          * @param string $name
 519          * @return bool
 520          */
 521         public function getCheck( $name ) {
 522                 # Checkboxes and buttons are only present when clicked
 523                 # Presence connotes truth, absence false
 524                 return $this->getVal( $name, null ) !== null;
 525         }
 526
 527         /**
 528          * Fetch a text string from the given array or return $default if it's not
 529          * set. Carriage returns are stripped from the text, and with some language
 530          * modules there is an input transliteration applied. This should generally
 531          * be used for form "<textarea>" and "<input>" fields. Used for
 532          * user-supplied freeform text input (for which input transformations may
 533          * be required - e.g.  Esperanto x-coding).
 534          *
 535          * @param string $name
 536          * @param string $default Optional
 537          * @return string
 538          */
 539         public function getText( $name, $default = '' ) {
 540                 global $wgContLang;
 541                 $val = $this->getVal( $name, $default );
 542                 return str_replace( "\r\n", "\n",
 543                         $wgContLang->recodeInput( $val ) );
 544         }
 545
 546         /**
 547          * Extracts the given named values into an array.
 548          * If no arguments are given, returns all input values.
 549          * No transformation is performed on the values.
 550          *
 551          * @return array
 552          */
 553         public function getValues() {
 554                 $names = func_get_args();
 555                 if ( count( $names ) == 0 ) {
 556                         $names = array_keys( $this->data );
 557                 }
 558
 559                 $retVal = array();
 560                 foreach ( $names as $name ) {
 561                         $value = $this->getGPCVal( $this->data, $name, null );
 562                         if ( !is_null( $value ) ) {
 563                                 $retVal[$name] = $value;
 564                         }
 565                 }
 566                 return $retVal;
 567         }
 568
 569         /**
 570          * Returns the names of all input values excluding those in $exclude.
 571          *
 572          * @param array $exclude
 573          * @return array
 574          */
 575         public function getValueNames( $exclude = array() ) {
 576                 return array_diff( array_keys( $this->getValues() ), $exclude );
 577         }
 578
 579         /**
 580          * Get the values passed in the query string.
 581          * No transformation is performed on the values.
 582          *
 583          * @return array
 584          */
 585         public function getQueryValues() {
 586                 return $_GET;
 587         }
 588
 589         /**
 590          * Return the contents of the Query with no decoding. Use when you need to
 591          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 592          *
 593          * @return string
 594          */
 595         public function getRawQueryString() {
 596                 return $_SERVER['QUERY_STRING'];
 597         }
 598
 599         /**
 600          * Return the contents of the POST with no decoding. Use when you need to
 601          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 602          *
 603          * @return string
 604          */
 605         public function getRawPostString() {
 606                 if ( !$this->wasPosted() ) {
 607                         return '';
 608                 }
 609                 return $this->getRawInput();
 610         }
 611
 612         /**
 613          * Return the raw request body, with no processing. Cached since some methods
 614          * disallow reading the stream more than once. As stated in the php docs, this
 615          * does not work with enctype="multipart/form-data".
 616          *
 617          * @return string
 618          */
 619         public function getRawInput() {
 620                 static $input = null;
 621                 if ( $input === null ) {
 622                         $input = file_get_contents( 'php://input' );
 623                 }
 624                 return $input;
 625         }
 626
 627         /**
 628          * Get the HTTP method used for this request.
 629          *
 630          * @return string
 631          */
 632         public function getMethod() {
 633                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 634         }
 635
 636         /**
 637          * Returns true if the present request was reached by a POST operation,
 638          * false otherwise (GET, HEAD, or command-line).
 639          *
 640          * Note that values retrieved by the object may come from the
 641          * GET URL etc even on a POST request.
 642          *
 643          * @return bool
 644          */
 645         public function wasPosted() {
 646                 return $this->getMethod() == 'POST';
 647         }
 648
 649         /**
 650          * Return the session for this request
 651          * @since 1.27
 652          * @note For performance, keep the session locally if you will be making
 653          *  much use of it instead of calling this method repeatedly.
 654          * @return MediaWiki\\Session\\Session
 655          */
 656         public function getSession() {
 657                 if ( $this->sessionId !== null ) {
 658                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 659                         if ( $session ) {
 660                                 return $session;
 661                         }
 662                 }
 663
 664                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 665                 $this->sessionId = $session->getSessionId();
 666                 return $session;
 667         }
 668
 669         /**
 670          * Set the session for this request
 671          * @since 1.27
 672          * @private For use by MediaWiki\\Session classes only
 673          * @param MediaWiki\\Session\\SessionId $sessionId
 674          */
 675         public function setSessionId( MediaWiki\Session\SessionId $sessionId ) {
 676                 $this->sessionId = $sessionId;
 677         }
 678
 679         /**
 680          * Returns true if the request has a persistent session.
 681          * This does not necessarily mean that the user is logged in!
 682          *
 683          * @deprecated since 1.27, use
 684          *  \\MediaWiki\\Session\\SessionManager::singleton()->getPersistedSessionId()
 685          *  instead.
 686          * @return bool
 687          */
 688         public function checkSessionCookie() {
 689                 wfDeprecated( __METHOD__, '1.27' );
 690                 return SessionManager::singleton()->getPersistedSessionId( $this ) !== null;
 691         }
 692
 693         /**
 694          * Get a cookie from the $_COOKIE jar
 695          *
 696          * @param string $key The name of the cookie
 697          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 698          * @param mixed $default What to return if the value isn't found
 699          * @return mixed Cookie value or $default if the cookie not set
 700          */
 701         public function getCookie( $key, $prefix = null, $default = null ) {
 702                 if ( $prefix === null ) {
 703                         global $wgCookiePrefix;
 704                         $prefix = $wgCookiePrefix;
 705                 }
 706                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 707         }
 708
 709         /**
 710          * Return the path and query string portion of the request URI.
 711          * This will be suitable for use as a relative link in HTML output.
 712          *
 713          * @throws MWException
 714          * @return string
 715          */
 716         public function getRequestURL() {
 717                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 718                         $base = $_SERVER['REQUEST_URI'];
 719                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 720                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 721                 ) {
 722                         // Probably IIS; doesn't set REQUEST_URI
 723                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 724                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 725                         $base = $_SERVER['SCRIPT_NAME'];
 726                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 727                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 728                         }
 729                 } else {
 730                         // This shouldn't happen!
 731                         throw new MWException( "Web server doesn't provide either " .
 732                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 733                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 734                 }
 735                 // User-agents should not send a fragment with the URI, but
 736                 // if they do, and the web server passes it on to us, we
 737                 // need to strip it or we get false-positive redirect loops
 738                 // or weird output URLs
 739                 $hash = strpos( $base, '#' );
 740                 if ( $hash !== false ) {
 741                         $base = substr( $base, 0, $hash );
 742                 }
 743
 744                 if ( $base[0] == '/' ) {
 745                         // More than one slash will look like it is protocol relative
 746                         return preg_replace( '!^/+!', '/', $base );
 747                 } else {
 748                         // We may get paths with a host prepended; strip it.
 749                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 750                 }
 751         }
 752
 753         /**
 754          * Return the request URI with the canonical service and hostname, path,
 755          * and query string. This will be suitable for use as an absolute link
 756          * in HTML or other output.
 757          *
 758          * If $wgServer is protocol-relative, this will return a fully
 759          * qualified URL with the protocol that was used for this request.
 760          *
 761          * @return string
 762          */
 763         public function getFullRequestURL() {
 764                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 765         }
 766
 767         /**
 768          * @param string $key
 769          * @param string $value
 770          * @return string
 771          */
 772         public function appendQueryValue( $key, $value ) {
 773                 return $this->appendQueryArray( array( $key => $value ) );
 774         }
 775
 776         /**
 777          * Appends or replaces value of query variables.
 778          *
 779          * @param array $array Array of values to replace/add to query
 780          * @return string
 781          */
 782         public function appendQueryArray( $array ) {
 783                 $newquery = $this->getQueryValues();
 784                 unset( $newquery['title'] );
 785                 $newquery = array_merge( $newquery, $array );
 786
 787                 return wfArrayToCgi( $newquery );
 788         }
 789
 790         /**
 791          * Check for limit and offset parameters on the input, and return sensible
 792          * defaults if not given. The limit must be positive and is capped at 5000.
 793          * Offset must be positive but is not capped.
 794          *
 795          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 796          * @param string $optionname To specify an option other than rclimit to pull from.
 797          * @return int[] First element is limit, second is offset
 798          */
 799         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 800                 global $wgUser;
 801
 802                 $limit = $this->getInt( 'limit', 0 );
 803                 if ( $limit < 0 ) {
 804                         $limit = 0;
 805                 }
 806                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 807                         $limit = $wgUser->getIntOption( $optionname );
 808                 }
 809                 if ( $limit <= 0 ) {
 810                         $limit = $deflimit;
 811                 }
 812                 if ( $limit > 5000 ) {
 813                         $limit = 5000; # We have *some* limits...
 814                 }
 815
 816                 $offset = $this->getInt( 'offset', 0 );
 817                 if ( $offset < 0 ) {
 818                         $offset = 0;
 819                 }
 820
 821                 return array( $limit, $offset );
 822         }
 823
 824         /**
 825          * Return the path to the temporary file where PHP has stored the upload.
 826          *
 827          * @param string $key
 828          * @return string|null String or null if no such file.
 829          */
 830         public function getFileTempname( $key ) {
 831                 $file = new WebRequestUpload( $this, $key );
 832                 return $file->getTempName();
 833         }
 834
 835         /**
 836          * Return the upload error or 0
 837          *
 838          * @param string $key
 839          * @return int
 840          */
 841         public function getUploadError( $key ) {
 842                 $file = new WebRequestUpload( $this, $key );
 843                 return $file->getError();
 844         }
 845
 846         /**
 847          * Return the original filename of the uploaded file, as reported by
 848          * the submitting user agent. HTML-style character entities are
 849          * interpreted and normalized to Unicode normalization form C, in part
 850          * to deal with weird input from Safari with non-ASCII filenames.
 851          *
 852          * Other than this the name is not verified for being a safe filename.
 853          *
 854          * @param string $key
 855          * @return string|null String or null if no such file.
 856          */
 857         public function getFileName( $key ) {
 858                 $file = new WebRequestUpload( $this, $key );
 859                 return $file->getName();
 860         }
 861
 862         /**
 863          * Return a WebRequestUpload object corresponding to the key
 864          *
 865          * @param string $key
 866          * @return WebRequestUpload
 867          */
 868         public function getUpload( $key ) {
 869                 return new WebRequestUpload( $this, $key );
 870         }
 871
 872         /**
 873          * Return a handle to WebResponse style object, for setting cookies,
 874          * headers and other stuff, for Request being worked on.
 875          *
 876          * @return WebResponse
 877          */
 878         public function response() {
 879                 /* Lazy initialization of response object for this request */
 880                 if ( !is_object( $this->response ) ) {
 881                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 882                         $this->response = new $class();
 883                 }
 884                 return $this->response;
 885         }
 886
 887         /**
 888          * Initialise the header list
 889          */
 890         protected function initHeaders() {
 891                 if ( count( $this->headers ) ) {
 892                         return;
 893                 }
 894
 895                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 896                 if ( $apacheHeaders ) {
 897                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 898                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 899                         }
 900                 } else {
 901                         foreach ( $_SERVER as $name => $value ) {
 902                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 903                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 904                                         $this->headers[$name] = $value;
 905                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 906                                         $this->headers['CONTENT-LENGTH'] = $value;
 907                                 }
 908                         }
 909                 }
 910         }
 911
 912         /**
 913          * Get an array containing all request headers
 914          *
 915          * @return array Mapping header name to its value
 916          */
 917         public function getAllHeaders() {
 918                 $this->initHeaders();
 919                 return $this->headers;
 920         }
 921
 922         /**
 923          * Get a request header, or false if it isn't set.
 924          *
 925          * @param string $name Case-insensitive header name
 926          * @param int $flags Bitwise combination of:
 927          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
 928          *                               of values, as described in RFC 2616 § 4.2.
 929          *                               (since 1.26).
 930          * @return string|array|bool False if header is unset; otherwise the
 931          *  header value(s) as either a string (the default) or an array, if
 932          *  WebRequest::GETHEADER_LIST flag was set.
 933          */
 934         public function getHeader( $name, $flags = 0 ) {
 935                 $this->initHeaders();
 936                 $name = strtoupper( $name );
 937                 if ( !isset( $this->headers[$name] ) ) {
 938                         return false;
 939                 }
 940                 $value = $this->headers[$name];
 941                 if ( $flags & self::GETHEADER_LIST ) {
 942                         $value = array_map( 'trim', explode( ',', $value ) );
 943                 }
 944                 return $value;
 945         }
 946
 947         /**
 948          * Get data from the session
 949          *
 950          * @note Prefer $this->getSession() instead if making multiple calls.
 951          * @param string $key Name of key in the session
 952          * @return mixed
 953          */
 954         public function getSessionData( $key ) {
 955                 return $this->getSession()->get( $key );
 956         }
 957
 958         /**
 959          * Set session data
 960          *
 961          * @note Prefer $this->getSession() instead if making multiple calls.
 962          * @param string $key Name of key in the session
 963          * @param mixed $data
 964          */
 965         public function setSessionData( $key, $data ) {
 966                 return $this->getSession()->set( $key, $data );
 967         }
 968
 969         /**
 970          * Check if Internet Explorer will detect an incorrect cache extension in
 971          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 972          * message or redirect to a safer URL. Returns true if the URL is OK, and
 973          * false if an error message has been shown and the request should be aborted.
 974          *
 975          * @param array $extWhitelist
 976          * @throws HttpError
 977          * @return bool
 978          */
 979         public function checkUrlExtension( $extWhitelist = array() ) {
 980                 $extWhitelist[] = 'php';
 981                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 982                         if ( !$this->wasPosted() ) {
 983                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 984                                         $this->getFullRequestURL(), $extWhitelist );
 985                                 if ( $newUrl !== false ) {
 986                                         $this->doSecurityRedirect( $newUrl );
 987                                         return false;
 988                                 }
 989                         }
 990                         throw new HttpError( 403,
 991                                 'Invalid file extension found in the path info or query string.' );
 992                 }
 993                 return true;
 994         }
 995
 996         /**
 997          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 998          * IE 6. Returns true if it was successful, false otherwise.
 999          *
1000          * @param string $url
1001          * @return bool
1002          */
1003         protected function doSecurityRedirect( $url ) {
1004                 header( 'Location: ' . $url );
1005                 header( 'Content-Type: text/html' );
1006                 $encUrl = htmlspecialchars( $url );
1007                 echo <<<HTML
1008 <html>
1009 <head>
1010 <title>Security redirect</title>
1011 </head>
1012 <body>
1013 <h1>Security redirect</h1>
1014 <p>
1015 We can't serve non-HTML content from the URL you have requested, because
1016 Internet Explorer would interpret it as an incorrect and potentially dangerous
1017 content type.</p>
1018 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1019 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1020 Explorer from seeing a bogus file extension.
1021 </p>
1022 </body>
1023 </html>
1024 HTML;
1025                 echo "\n";
1026                 return true;
1027         }
1028
1029         /**
1030          * Parse the Accept-Language header sent by the client into an array
1031          *
1032          * @return array Array( languageCode => q-value ) sorted by q-value in
1033          *   descending order then appearing time in the header in ascending order.
1034          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1035          * This is aligned with rfc2616 section 14.4
1036          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1037          */
1038         public function getAcceptLang() {
1039                 // Modified version of code found at
1040                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1041                 $acceptLang = $this->getHeader( 'Accept-Language' );
1042                 if ( !$acceptLang ) {
1043                         return array();
1044                 }
1045
1046                 // Return the language codes in lower case
1047                 $acceptLang = strtolower( $acceptLang );
1048
1049                 // Break up string into pieces (languages and q factors)
1050                 $lang_parse = null;
1051                 preg_match_all(
1052                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1053                         $acceptLang,
1054                         $lang_parse
1055                 );
1056
1057                 if ( !count( $lang_parse[1] ) ) {
1058                         return array();
1059                 }
1060
1061                 $langcodes = $lang_parse[1];
1062                 $qvalues = $lang_parse[4];
1063                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1064
1065                 // Set default q factor to 1
1066                 foreach ( $indices as $index ) {
1067                         if ( $qvalues[$index] === '' ) {
1068                                 $qvalues[$index] = 1;
1069                         } elseif ( $qvalues[$index] == 0 ) {
1070                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1071                         }
1072                 }
1073
1074                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1075                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1076
1077                 // Create a list like "en" => 0.8
1078                 $langs = array_combine( $langcodes, $qvalues );
1079
1080                 return $langs;
1081         }
1082
1083         /**
1084          * Fetch the raw IP from the request
1085          *
1086          * @since 1.19
1087          *
1088          * @throws MWException
1089          * @return string
1090          */
1091         protected function getRawIP() {
1092                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1093                         return null;
1094                 }
1095
1096                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1097                         throw new MWException( __METHOD__
1098                                 . " : Could not determine the remote IP address due to multiple values." );
1099                 } else {
1100                         $ipchain = $_SERVER['REMOTE_ADDR'];
1101                 }
1102
1103                 return IP::canonicalize( $ipchain );
1104         }
1105
1106         /**
1107          * Work out the IP address based on various globals
1108          * For trusted proxies, use the XFF client IP (first of the chain)
1109          *
1110          * @since 1.19
1111          *
1112          * @throws MWException
1113          * @return string
1114          */
1115         public function getIP() {
1116                 global $wgUsePrivateIPs;
1117
1118                 # Return cached result
1119                 if ( $this->ip !== null ) {
1120                         return $this->ip;
1121                 }
1122
1123                 # collect the originating ips
1124                 $ip = $this->getRawIP();
1125                 if ( !$ip ) {
1126                         throw new MWException( 'Unable to determine IP.' );
1127                 }
1128
1129                 # Append XFF
1130                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1131                 if ( $forwardedFor !== false ) {
1132                         $isConfigured = IP::isConfiguredProxy( $ip );
1133                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1134                         $ipchain = array_reverse( $ipchain );
1135                         array_unshift( $ipchain, $ip );
1136
1137                         # Step through XFF list and find the last address in the list which is a
1138                         # trusted server. Set $ip to the IP address given by that trusted server,
1139                         # unless the address is not sensible (e.g. private). However, prefer private
1140                         # IP addresses over proxy servers controlled by this site (more sensible).
1141                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1142                         foreach ( $ipchain as $i => $curIP ) {
1143                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1144                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1145                                         || !IP::isTrustedProxy( $curIP )
1146                                 ) {
1147                                         break; // IP is not valid/trusted or does not point to anything
1148                                 }
1149                                 if (
1150                                         IP::isPublic( $ipchain[$i + 1] ) ||
1151                                         $wgUsePrivateIPs ||
1152                                         IP::isConfiguredProxy( $curIP ) // bug 48919; treat IP as sane
1153                                 ) {
1154                                         // Follow the next IP according to the proxy
1155                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1156                                         if ( !$nextIP && $isConfigured ) {
1157                                                 // We have not yet made it past CDN/proxy servers of this site,
1158                                                 // so either they are misconfigured or there is some IP spoofing.
1159                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1160                                         }
1161                                         $ip = $nextIP;
1162                                         // keep traversing the chain
1163                                         continue;
1164                                 }
1165                                 break;
1166                         }
1167                 }
1168
1169                 # Allow extensions to improve our guess
1170                 Hooks::run( 'GetIP', array( &$ip ) );
1171
1172                 if ( !$ip ) {
1173                         throw new MWException( "Unable to determine IP." );
1174                 }
1175
1176                 wfDebug( "IP: $ip\n" );
1177                 $this->ip = $ip;
1178                 return $ip;
1179         }
1180
1181         /**
1182          * @param string $ip
1183          * @return void
1184          * @since 1.21
1185          */
1186         public function setIP( $ip ) {
1187                 $this->ip = $ip;
1188         }
1189 }