includes/WebRequest.php

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