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