includes/WebRequest.php

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