includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 use MediaWiki\Session\Session;
  27 use MediaWiki\Session\SessionId;
  28 use MediaWiki\Session\SessionManager;
  29
  30 /**
  31  * The WebRequest class encapsulates getting at data passed in the
  32  * URL or via a POSTed form stripping illegal input characters and
  33  * normalizing Unicode sequences.
  34  *
  35  * @ingroup HTTP
  36  */
  37 class WebRequest {
  38         protected $data, $headers = [];
  39
  40         /**
  41          * Flag to make WebRequest::getHeader return an array of values.
  42          * @since 1.26
  43          */
  44         const GETHEADER_LIST = 1;
  45
  46         /**
  47          * The unique request ID.
  48          * @var string
  49          */
  50         private static $reqId;
  51
  52         /**
  53          * Lazy-init response object
  54          * @var WebResponse
  55          */
  56         private $response;
  57
  58         /**
  59          * Cached client IP address
  60          * @var string
  61          */
  62         private $ip;
  63
  64         /**
  65          * The timestamp of the start of the request, with microsecond precision.
  66          * @var float
  67          */
  68         protected $requestTime;
  69
  70         /**
  71          * Cached URL protocol
  72          * @var string
  73          */
  74         protected $protocol;
  75
  76         /**
  77          * @var SessionId|null Session ID to use for this
  78          *  request. We can't save the session directly due to reference cycles not
  79          *  working too well (slow GC in Zend and never collected in HHVM).
  80          */
  81         protected $sessionId = null;
  82
  83         /** @var bool Whether this HTTP request is "safe" (even if it is an HTTP post) */
  84         protected $markedAsSafe = false;
  85
  86         public function __construct() {
  87                 $this->requestTime = isset( $_SERVER['REQUEST_TIME_FLOAT'] )
  88                         ? $_SERVER['REQUEST_TIME_FLOAT'] : microtime( true );
  89
  90                 // POST overrides GET data
  91                 // We don't use $_REQUEST here to avoid interference from cookies...
  92                 $this->data = $_POST + $_GET;
  93         }
  94
  95         /**
  96          * Extract relevant query arguments from the http request uri's path
  97          * to be merged with the normal php provided query arguments.
  98          * Tries to use the REQUEST_URI data if available and parses it
  99          * according to the wiki's configuration looking for any known pattern.
 100          *
 101          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
 102          * provided by the server if any and use that to set a 'title' parameter.
 103          *
 104          * @param string $want If this is not 'all', then the function
 105          * will return an empty array if it determines that the URL is
 106          * inside a rewrite path.
 107          *
 108          * @return array Any query arguments found in path matches.
 109          */
 110         public static function getPathInfo( $want = 'all' ) {
 111                 global $wgUsePathInfo;
 112                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
 113                 // And also by Apache 2.x, double slashes are converted to single slashes.
 114                 // So we will use REQUEST_URI if possible.
 115                 $matches = [];
 116                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
 117                         // Slurp out the path portion to examine...
 118                         $url = $_SERVER['REQUEST_URI'];
 119                         if ( !preg_match( '!^https?://!', $url ) ) {
 120                                 $url = 'http://unused' . $url;
 121                         }
 122                         MediaWiki\suppressWarnings();
 123                         $a = parse_url( $url );
 124                         MediaWiki\restoreWarnings();
 125                         if ( $a ) {
 126                                 $path = isset( $a['path'] ) ? $a['path'] : '';
 127
 128                                 global $wgScript;
 129                                 if ( $path == $wgScript && $want !== 'all' ) {
 130                                         // Script inside a rewrite path?
 131                                         // Abort to keep from breaking...
 132                                         return $matches;
 133                                 }
 134
 135                                 $router = new PathRouter;
 136
 137                                 // Raw PATH_INFO style
 138                                 $router->add( "$wgScript/$1" );
 139
 140                                 if ( isset( $_SERVER['SCRIPT_NAME'] )
 141                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] )
 142                                 ) {
 143                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 144                                         # But we do have some other .php files such as img_auth.php
 145                                         # Don't let root article paths clober the parsing for them
 146                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 147                                 }
 148
 149                                 global $wgArticlePath;
 150                                 if ( $wgArticlePath ) {
 151                                         $router->add( $wgArticlePath );
 152                                 }
 153
 154                                 global $wgActionPaths;
 155                                 if ( $wgActionPaths ) {
 156                                         $router->add( $wgActionPaths, [ 'action' => '$key' ] );
 157                                 }
 158
 159                                 global $wgVariantArticlePath, $wgContLang;
 160                                 if ( $wgVariantArticlePath ) {
 161                                         $router->add( $wgVariantArticlePath,
 162                                                 [ 'variant' => '$2' ],
 163                                                 [ '$2' => $wgContLang->getVariants() ]
 164                                         );
 165                                 }
 166
 167                                 Hooks::run( 'WebRequestPathInfoRouter', [ $router ] );
 168
 169                                 $matches = $router->parse( $path );
 170                         }
 171                 } elseif ( $wgUsePathInfo ) {
 172                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 173                                 // Mangled PATH_INFO
 174                                 // http://bugs.php.net/bug.php?id=31892
 175                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 176                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 177
 178                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && $_SERVER['PATH_INFO'] != '' ) {
 179                                 // Regular old PATH_INFO yay
 180                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 181                         }
 182                 }
 183
 184                 return $matches;
 185         }
 186
 187         /**
 188          * Work out an appropriate URL prefix containing scheme and host, based on
 189          * information detected from $_SERVER
 190          *
 191          * @return string
 192          */
 193         public static function detectServer() {
 194                 global $wgAssumeProxiesUseDefaultProtocolPorts;
 195
 196                 $proto = self::detectProtocol();
 197                 $stdPort = $proto === 'https' ? 443 : 80;
 198
 199                 $varNames = [ 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' ];
 200                 $host = 'localhost';
 201                 $port = $stdPort;
 202                 foreach ( $varNames as $varName ) {
 203                         if ( !isset( $_SERVER[$varName] ) ) {
 204                                 continue;
 205                         }
 206
 207                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 208                         if ( !$parts ) {
 209                                 // Invalid, do not use
 210                                 continue;
 211                         }
 212
 213                         $host = $parts[0];
 214                         if ( $wgAssumeProxiesUseDefaultProtocolPorts && isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) ) {
 215                                 // Bug 70021: Assume that upstream proxy is running on the default
 216                                 // port based on the protocol. We have no reliable way to determine
 217                                 // the actual port in use upstream.
 218                                 $port = $stdPort;
 219                         } elseif ( $parts[1] === false ) {
 220                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 221                                         $port = $_SERVER['SERVER_PORT'];
 222                                 } // else leave it as $stdPort
 223                         } else {
 224                                 $port = $parts[1];
 225                         }
 226                         break;
 227                 }
 228
 229                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 230         }
 231
 232         /**
 233          * Detect the protocol from $_SERVER.
 234          * This is for use prior to Setup.php, when no WebRequest object is available.
 235          * At other times, use the non-static function getProtocol().
 236          *
 237          * @return array
 238          */
 239         public static function detectProtocol() {
 240                 if ( ( !empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) ||
 241                         ( isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) &&
 242                         $_SERVER['HTTP_X_FORWARDED_PROTO'] === 'https' ) ) {
 243                         return 'https';
 244                 } else {
 245                         return 'http';
 246                 }
 247         }
 248
 249         /**
 250          * Get the number of seconds to have elapsed since request start,
 251          * in fractional seconds, with microsecond resolution.
 252          *
 253          * @return float
 254          * @since 1.25
 255          */
 256         public function getElapsedTime() {
 257                 return microtime( true ) - $this->requestTime;
 258         }
 259
 260         /**
 261          * Get the unique request ID.
 262          * This is either the value of the UNIQUE_ID envvar (if present) or a
 263          * randomly-generated 24-character string.
 264          *
 265          * @return string
 266          * @since 1.27
 267          */
 268         public static function getRequestId() {
 269                 if ( !self::$reqId ) {
 270                         self::$reqId = isset( $_SERVER['UNIQUE_ID'] )
 271                                 ? $_SERVER['UNIQUE_ID'] : wfRandomString( 24 );
 272                 }
 273
 274                 return self::$reqId;
 275         }
 276
 277         /**
 278          * Override the unique request ID. This is for sub-requests, such as jobs,
 279          * that wish to use the same id but are not part of the same execution context.
 280          *
 281          * @param string $id
 282          * @since 1.27
 283          */
 284         public static function overrideRequestId( $id ) {
 285                 self::$reqId = $id;
 286         }
 287
 288         /**
 289          * Get the current URL protocol (http or https)
 290          * @return string
 291          */
 292         public function getProtocol() {
 293                 if ( $this->protocol === null ) {
 294                         $this->protocol = self::detectProtocol();
 295                 }
 296                 return $this->protocol;
 297         }
 298
 299         /**
 300          * Check for title, action, and/or variant data in the URL
 301          * and interpolate it into the GET variables.
 302          * This should only be run after $wgContLang is available,
 303          * as we may need the list of language variants to determine
 304          * available variant URLs.
 305          */
 306         public function interpolateTitle() {
 307                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 308                 if ( defined( 'MW_API' ) ) {
 309                         return;
 310                 }
 311
 312                 $matches = self::getPathInfo( 'title' );
 313                 foreach ( $matches as $key => $val ) {
 314                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 315                 }
 316         }
 317
 318         /**
 319          * URL rewriting function; tries to extract page title and,
 320          * optionally, one other fixed parameter value from a URL path.
 321          *
 322          * @param string $path The URL path given from the client
 323          * @param array $bases One or more URLs, optionally with $1 at the end
 324          * @param string|bool $key If provided, the matching key in $bases will be
 325          *    passed on as the value of this URL parameter
 326          * @return array Array of URL variables to interpolate; empty if no match
 327          */
 328         static function extractTitle( $path, $bases, $key = false ) {
 329                 foreach ( (array)$bases as $keyValue => $base ) {
 330                         // Find the part after $wgArticlePath
 331                         $base = str_replace( '$1', '', $base );
 332                         $baseLen = strlen( $base );
 333                         if ( substr( $path, 0, $baseLen ) == $base ) {
 334                                 $raw = substr( $path, $baseLen );
 335                                 if ( $raw !== '' ) {
 336                                         $matches = [ 'title' => rawurldecode( $raw ) ];
 337                                         if ( $key ) {
 338                                                 $matches[$key] = $keyValue;
 339                                         }
 340                                         return $matches;
 341                                 }
 342                         }
 343                 }
 344                 return [];
 345         }
 346
 347         /**
 348          * Recursively normalizes UTF-8 strings in the given array.
 349          *
 350          * @param string|array $data
 351          * @return array|string Cleaned-up version of the given
 352          * @private
 353          */
 354         function normalizeUnicode( $data ) {
 355                 if ( is_array( $data ) ) {
 356                         foreach ( $data as $key => $val ) {
 357                                 $data[$key] = $this->normalizeUnicode( $val );
 358                         }
 359                 } else {
 360                         global $wgContLang;
 361                         $data = isset( $wgContLang ) ?
 362                                 $wgContLang->normalize( $data ) :
 363                                 UtfNormal\Validator::cleanUp( $data );
 364                 }
 365                 return $data;
 366         }
 367
 368         /**
 369          * Fetch a value from the given array or return $default if it's not set.
 370          *
 371          * @param array $arr
 372          * @param string $name
 373          * @param mixed $default
 374          * @return mixed
 375          */
 376         private function getGPCVal( $arr, $name, $default ) {
 377                 # PHP is so nice to not touch input data, except sometimes:
 378                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 379                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 380                 $name = strtr( $name, '.', '_' );
 381                 if ( isset( $arr[$name] ) ) {
 382                         global $wgContLang;
 383                         $data = $arr[$name];
 384                         if ( isset( $_GET[$name] ) && !is_array( $data ) ) {
 385                                 # Check for alternate/legacy character encoding.
 386                                 if ( isset( $wgContLang ) ) {
 387                                         $data = $wgContLang->checkTitleEncoding( $data );
 388                                 }
 389                         }
 390                         $data = $this->normalizeUnicode( $data );
 391                         return $data;
 392                 } else {
 393                         return $default;
 394                 }
 395         }
 396
 397         /**
 398          * Fetch a scalar from the input or return $default if it's not set.
 399          * Returns a string. Arrays are discarded. Useful for
 400          * non-freeform text inputs (e.g. predefined internal text keys
 401          * selected by a drop-down menu). For freeform input, see getText().
 402          *
 403          * @param string $name
 404          * @param string $default Optional default (or null)
 405          * @return string
 406          */
 407         public function getVal( $name, $default = null ) {
 408                 $val = $this->getGPCVal( $this->data, $name, $default );
 409                 if ( is_array( $val ) ) {
 410                         $val = $default;
 411                 }
 412                 if ( is_null( $val ) ) {
 413                         return $val;
 414                 } else {
 415                         return (string)$val;
 416                 }
 417         }
 418
 419         /**
 420          * Set an arbitrary value into our get/post data.
 421          *
 422          * @param string $key Key name to use
 423          * @param mixed $value Value to set
 424          * @return mixed Old value if one was present, null otherwise
 425          */
 426         public function setVal( $key, $value ) {
 427                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 428                 $this->data[$key] = $value;
 429                 return $ret;
 430         }
 431
 432         /**
 433          * Unset an arbitrary value from our get/post data.
 434          *
 435          * @param string $key Key name to use
 436          * @return mixed Old value if one was present, null otherwise
 437          */
 438         public function unsetVal( $key ) {
 439                 if ( !isset( $this->data[$key] ) ) {
 440                         $ret = null;
 441                 } else {
 442                         $ret = $this->data[$key];
 443                         unset( $this->data[$key] );
 444                 }
 445                 return $ret;
 446         }
 447
 448         /**
 449          * Fetch an array from the input or return $default if it's not set.
 450          * If source was scalar, will return an array with a single element.
 451          * If no source and no default, returns null.
 452          *
 453          * @param string $name
 454          * @param array $default Optional default (or null)
 455          * @return array
 456          */
 457         public function getArray( $name, $default = null ) {
 458                 $val = $this->getGPCVal( $this->data, $name, $default );
 459                 if ( is_null( $val ) ) {
 460                         return null;
 461                 } else {
 462                         return (array)$val;
 463                 }
 464         }
 465
 466         /**
 467          * Fetch an array of integers, or return $default if it's not set.
 468          * If source was scalar, will return an array with a single element.
 469          * If no source and no default, returns null.
 470          * If an array is returned, contents are guaranteed to be integers.
 471          *
 472          * @param string $name
 473          * @param array $default Option default (or null)
 474          * @return array Array of ints
 475          */
 476         public function getIntArray( $name, $default = null ) {
 477                 $val = $this->getArray( $name, $default );
 478                 if ( is_array( $val ) ) {
 479                         $val = array_map( 'intval', $val );
 480                 }
 481                 return $val;
 482         }
 483
 484         /**
 485          * Fetch an integer value from the input or return $default if not set.
 486          * Guaranteed to return an integer; non-numeric input will typically
 487          * return 0.
 488          *
 489          * @param string $name
 490          * @param int $default
 491          * @return int
 492          */
 493         public function getInt( $name, $default = 0 ) {
 494                 return intval( $this->getVal( $name, $default ) );
 495         }
 496
 497         /**
 498          * Fetch an integer value from the input or return null if empty.
 499          * Guaranteed to return an integer or null; non-numeric input will
 500          * typically return null.
 501          *
 502          * @param string $name
 503          * @return int|null
 504          */
 505         public function getIntOrNull( $name ) {
 506                 $val = $this->getVal( $name );
 507                 return is_numeric( $val )
 508                         ? intval( $val )
 509                         : null;
 510         }
 511
 512         /**
 513          * Fetch a floating point value from the input or return $default if not set.
 514          * Guaranteed to return a float; non-numeric input will typically
 515          * return 0.
 516          *
 517          * @since 1.23
 518          * @param string $name
 519          * @param float $default
 520          * @return float
 521          */
 522         public function getFloat( $name, $default = 0.0 ) {
 523                 return floatval( $this->getVal( $name, $default ) );
 524         }
 525
 526         /**
 527          * Fetch a boolean value from the input or return $default if not set.
 528          * Guaranteed to return true or false, with normal PHP semantics for
 529          * boolean interpretation of strings.
 530          *
 531          * @param string $name
 532          * @param bool $default
 533          * @return bool
 534          */
 535         public function getBool( $name, $default = false ) {
 536                 return (bool)$this->getVal( $name, $default );
 537         }
 538
 539         /**
 540          * Fetch a boolean value from the input or return $default if not set.
 541          * Unlike getBool, the string "false" will result in boolean false, which is
 542          * useful when interpreting information sent from JavaScript.
 543          *
 544          * @param string $name
 545          * @param bool $default
 546          * @return bool
 547          */
 548         public function getFuzzyBool( $name, $default = false ) {
 549                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 550         }
 551
 552         /**
 553          * Return true if the named value is set in the input, whatever that
 554          * value is (even "0"). Return false if the named value is not set.
 555          * Example use is checking for the presence of check boxes in forms.
 556          *
 557          * @param string $name
 558          * @return bool
 559          */
 560         public function getCheck( $name ) {
 561                 # Checkboxes and buttons are only present when clicked
 562                 # Presence connotes truth, absence false
 563                 return $this->getVal( $name, null ) !== null;
 564         }
 565
 566         /**
 567          * Fetch a text string from the given array or return $default if it's not
 568          * set. Carriage returns are stripped from the text, and with some language
 569          * modules there is an input transliteration applied. This should generally
 570          * be used for form "<textarea>" and "<input>" fields, and for
 571          * user-supplied freeform text input.
 572          *
 573          * @param string $name
 574          * @param string $default Optional
 575          * @return string
 576          */
 577         public function getText( $name, $default = '' ) {
 578                 $val = $this->getVal( $name, $default );
 579                 return str_replace( "\r\n", "\n", $val );
 580         }
 581
 582         /**
 583          * Extracts the given named values into an array.
 584          * If no arguments are given, returns all input values.
 585          * No transformation is performed on the values.
 586          *
 587          * @return array
 588          */
 589         public function getValues() {
 590                 $names = func_get_args();
 591                 if ( count( $names ) == 0 ) {
 592                         $names = array_keys( $this->data );
 593                 }
 594
 595                 $retVal = [];
 596                 foreach ( $names as $name ) {
 597                         $value = $this->getGPCVal( $this->data, $name, null );
 598                         if ( !is_null( $value ) ) {
 599                                 $retVal[$name] = $value;
 600                         }
 601                 }
 602                 return $retVal;
 603         }
 604
 605         /**
 606          * Returns the names of all input values excluding those in $exclude.
 607          *
 608          * @param array $exclude
 609          * @return array
 610          */
 611         public function getValueNames( $exclude = [] ) {
 612                 return array_diff( array_keys( $this->getValues() ), $exclude );
 613         }
 614
 615         /**
 616          * Get the values passed in the query string.
 617          * No transformation is performed on the values.
 618          *
 619          * @return array
 620          */
 621         public function getQueryValues() {
 622                 return $_GET;
 623         }
 624
 625         /**
 626          * Return the contents of the Query with no decoding. Use when you need to
 627          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 628          *
 629          * @return string
 630          */
 631         public function getRawQueryString() {
 632                 return $_SERVER['QUERY_STRING'];
 633         }
 634
 635         /**
 636          * Return the contents of the POST with no decoding. Use when you need to
 637          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 638          *
 639          * @return string
 640          */
 641         public function getRawPostString() {
 642                 if ( !$this->wasPosted() ) {
 643                         return '';
 644                 }
 645                 return $this->getRawInput();
 646         }
 647
 648         /**
 649          * Return the raw request body, with no processing. Cached since some methods
 650          * disallow reading the stream more than once. As stated in the php docs, this
 651          * does not work with enctype="multipart/form-data".
 652          *
 653          * @return string
 654          */
 655         public function getRawInput() {
 656                 static $input = null;
 657                 if ( $input === null ) {
 658                         $input = file_get_contents( 'php://input' );
 659                 }
 660                 return $input;
 661         }
 662
 663         /**
 664          * Get the HTTP method used for this request.
 665          *
 666          * @return string
 667          */
 668         public function getMethod() {
 669                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 670         }
 671
 672         /**
 673          * Returns true if the present request was reached by a POST operation,
 674          * false otherwise (GET, HEAD, or command-line).
 675          *
 676          * Note that values retrieved by the object may come from the
 677          * GET URL etc even on a POST request.
 678          *
 679          * @return bool
 680          */
 681         public function wasPosted() {
 682                 return $this->getMethod() == 'POST';
 683         }
 684
 685         /**
 686          * Return the session for this request
 687          * @since 1.27
 688          * @note For performance, keep the session locally if you will be making
 689          *  much use of it instead of calling this method repeatedly.
 690          * @return Session
 691          */
 692         public function getSession() {
 693                 if ( $this->sessionId !== null ) {
 694                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 695                         if ( $session ) {
 696                                 return $session;
 697                         }
 698                 }
 699
 700                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 701                 $this->sessionId = $session->getSessionId();
 702                 return $session;
 703         }
 704
 705         /**
 706          * Set the session for this request
 707          * @since 1.27
 708          * @private For use by MediaWiki\Session classes only
 709          * @param SessionId $sessionId
 710          */
 711         public function setSessionId( SessionId $sessionId ) {
 712                 $this->sessionId = $sessionId;
 713         }
 714
 715         /**
 716          * Get the session id for this request, if any
 717          * @since 1.27
 718          * @private For use by MediaWiki\Session classes only
 719          * @return SessionId|null
 720          */
 721         public function getSessionId() {
 722                 return $this->sessionId;
 723         }
 724
 725         /**
 726          * Returns true if the request has a persistent session.
 727          * This does not necessarily mean that the user is logged in!
 728          *
 729          * @deprecated since 1.27, use
 730          *  \MediaWiki\Session\SessionManager::singleton()->getPersistedSessionId()
 731          *  instead.
 732          * @return bool
 733          */
 734         public function checkSessionCookie() {
 735                 global $wgInitialSessionId;
 736                 wfDeprecated( __METHOD__, '1.27' );
 737                 return $wgInitialSessionId !== null &&
 738                         $this->getSession()->getId() === (string)$wgInitialSessionId;
 739         }
 740
 741         /**
 742          * Get a cookie from the $_COOKIE jar
 743          *
 744          * @param string $key The name of the cookie
 745          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 746          * @param mixed $default What to return if the value isn't found
 747          * @return mixed Cookie value or $default if the cookie not set
 748          */
 749         public function getCookie( $key, $prefix = null, $default = null ) {
 750                 if ( $prefix === null ) {
 751                         global $wgCookiePrefix;
 752                         $prefix = $wgCookiePrefix;
 753                 }
 754                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 755         }
 756
 757         /**
 758          * Return the path and query string portion of the main request URI.
 759          * This will be suitable for use as a relative link in HTML output.
 760          *
 761          * @throws MWException
 762          * @return string
 763          */
 764         public static function getGlobalRequestURL() {
 765                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 766                         $base = $_SERVER['REQUEST_URI'];
 767                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 768                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 769                 ) {
 770                         // Probably IIS; doesn't set REQUEST_URI
 771                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 772                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 773                         $base = $_SERVER['SCRIPT_NAME'];
 774                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 775                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 776                         }
 777                 } else {
 778                         // This shouldn't happen!
 779                         throw new MWException( "Web server doesn't provide either " .
 780                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 781                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 782                 }
 783                 // User-agents should not send a fragment with the URI, but
 784                 // if they do, and the web server passes it on to us, we
 785                 // need to strip it or we get false-positive redirect loops
 786                 // or weird output URLs
 787                 $hash = strpos( $base, '#' );
 788                 if ( $hash !== false ) {
 789                         $base = substr( $base, 0, $hash );
 790                 }
 791
 792                 if ( $base[0] == '/' ) {
 793                         // More than one slash will look like it is protocol relative
 794                         return preg_replace( '!^/+!', '/', $base );
 795                 } else {
 796                         // We may get paths with a host prepended; strip it.
 797                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 798                 }
 799         }
 800
 801         /**
 802          * Return the path and query string portion of the request URI.
 803          * This will be suitable for use as a relative link in HTML output.
 804          *
 805          * @throws MWException
 806          * @return string
 807          */
 808         public function getRequestURL() {
 809                 return self::getGlobalRequestURL();
 810         }
 811
 812         /**
 813          * Return the request URI with the canonical service and hostname, path,
 814          * and query string. This will be suitable for use as an absolute link
 815          * in HTML or other output.
 816          *
 817          * If $wgServer is protocol-relative, this will return a fully
 818          * qualified URL with the protocol that was used for this request.
 819          *
 820          * @return string
 821          */
 822         public function getFullRequestURL() {
 823                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 824         }
 825
 826         /**
 827          * @param string $key
 828          * @param string $value
 829          * @return string
 830          */
 831         public function appendQueryValue( $key, $value ) {
 832                 return $this->appendQueryArray( [ $key => $value ] );
 833         }
 834
 835         /**
 836          * Appends or replaces value of query variables.
 837          *
 838          * @param array $array Array of values to replace/add to query
 839          * @return string
 840          */
 841         public function appendQueryArray( $array ) {
 842                 $newquery = $this->getQueryValues();
 843                 unset( $newquery['title'] );
 844                 $newquery = array_merge( $newquery, $array );
 845
 846                 return wfArrayToCgi( $newquery );
 847         }
 848
 849         /**
 850          * Check for limit and offset parameters on the input, and return sensible
 851          * defaults if not given. The limit must be positive and is capped at 5000.
 852          * Offset must be positive but is not capped.
 853          *
 854          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 855          * @param string $optionname To specify an option other than rclimit to pull from.
 856          * @return int[] First element is limit, second is offset
 857          */
 858         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 859                 global $wgUser;
 860
 861                 $limit = $this->getInt( 'limit', 0 );
 862                 if ( $limit < 0 ) {
 863                         $limit = 0;
 864                 }
 865                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 866                         $limit = $wgUser->getIntOption( $optionname );
 867                 }
 868                 if ( $limit <= 0 ) {
 869                         $limit = $deflimit;
 870                 }
 871                 if ( $limit > 5000 ) {
 872                         $limit = 5000; # We have *some* limits...
 873                 }
 874
 875                 $offset = $this->getInt( 'offset', 0 );
 876                 if ( $offset < 0 ) {
 877                         $offset = 0;
 878                 }
 879
 880                 return [ $limit, $offset ];
 881         }
 882
 883         /**
 884          * Return the path to the temporary file where PHP has stored the upload.
 885          *
 886          * @param string $key
 887          * @return string|null String or null if no such file.
 888          */
 889         public function getFileTempname( $key ) {
 890                 $file = new WebRequestUpload( $this, $key );
 891                 return $file->getTempName();
 892         }
 893
 894         /**
 895          * Return the upload error or 0
 896          *
 897          * @param string $key
 898          * @return int
 899          */
 900         public function getUploadError( $key ) {
 901                 $file = new WebRequestUpload( $this, $key );
 902                 return $file->getError();
 903         }
 904
 905         /**
 906          * Return the original filename of the uploaded file, as reported by
 907          * the submitting user agent. HTML-style character entities are
 908          * interpreted and normalized to Unicode normalization form C, in part
 909          * to deal with weird input from Safari with non-ASCII filenames.
 910          *
 911          * Other than this the name is not verified for being a safe filename.
 912          *
 913          * @param string $key
 914          * @return string|null String or null if no such file.
 915          */
 916         public function getFileName( $key ) {
 917                 $file = new WebRequestUpload( $this, $key );
 918                 return $file->getName();
 919         }
 920
 921         /**
 922          * Return a WebRequestUpload object corresponding to the key
 923          *
 924          * @param string $key
 925          * @return WebRequestUpload
 926          */
 927         public function getUpload( $key ) {
 928                 return new WebRequestUpload( $this, $key );
 929         }
 930
 931         /**
 932          * Return a handle to WebResponse style object, for setting cookies,
 933          * headers and other stuff, for Request being worked on.
 934          *
 935          * @return WebResponse
 936          */
 937         public function response() {
 938                 /* Lazy initialization of response object for this request */
 939                 if ( !is_object( $this->response ) ) {
 940                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 941                         $this->response = new $class();
 942                 }
 943                 return $this->response;
 944         }
 945
 946         /**
 947          * Initialise the header list
 948          */
 949         protected function initHeaders() {
 950                 if ( count( $this->headers ) ) {
 951                         return;
 952                 }
 953
 954                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 955                 if ( $apacheHeaders ) {
 956                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 957                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 958                         }
 959                 } else {
 960                         foreach ( $_SERVER as $name => $value ) {
 961                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 962                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 963                                         $this->headers[$name] = $value;
 964                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 965                                         $this->headers['CONTENT-LENGTH'] = $value;
 966                                 }
 967                         }
 968                 }
 969         }
 970
 971         /**
 972          * Get an array containing all request headers
 973          *
 974          * @return array Mapping header name to its value
 975          */
 976         public function getAllHeaders() {
 977                 $this->initHeaders();
 978                 return $this->headers;
 979         }
 980
 981         /**
 982          * Get a request header, or false if it isn't set.
 983          *
 984          * @param string $name Case-insensitive header name
 985          * @param int $flags Bitwise combination of:
 986          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
 987          *                               of values, as described in RFC 2616 § 4.2.
 988          *                               (since 1.26).
 989          * @return string|array|bool False if header is unset; otherwise the
 990          *  header value(s) as either a string (the default) or an array, if
 991          *  WebRequest::GETHEADER_LIST flag was set.
 992          */
 993         public function getHeader( $name, $flags = 0 ) {
 994                 $this->initHeaders();
 995                 $name = strtoupper( $name );
 996                 if ( !isset( $this->headers[$name] ) ) {
 997                         return false;
 998                 }
 999                 $value = $this->headers[$name];
1000                 if ( $flags & self::GETHEADER_LIST ) {
1001                         $value = array_map( 'trim', explode( ',', $value ) );
1002                 }
1003                 return $value;
1004         }
1005
1006         /**
1007          * Get data from the session
1008          *
1009          * @note Prefer $this->getSession() instead if making multiple calls.
1010          * @param string $key Name of key in the session
1011          * @return mixed
1012          */
1013         public function getSessionData( $key ) {
1014                 return $this->getSession()->get( $key );
1015         }
1016
1017         /**
1018          * Set session data
1019          *
1020          * @note Prefer $this->getSession() instead if making multiple calls.
1021          * @param string $key Name of key in the session
1022          * @param mixed $data
1023          */
1024         public function setSessionData( $key, $data ) {
1025                 $this->getSession()->set( $key, $data );
1026         }
1027
1028         /**
1029          * Check if Internet Explorer will detect an incorrect cache extension in
1030          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1031          * message or redirect to a safer URL. Returns true if the URL is OK, and
1032          * false if an error message has been shown and the request should be aborted.
1033          *
1034          * @param array $extWhitelist
1035          * @throws HttpError
1036          * @return bool
1037          */
1038         public function checkUrlExtension( $extWhitelist = [] ) {
1039                 $extWhitelist[] = 'php';
1040                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1041                         if ( !$this->wasPosted() ) {
1042                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1043                                         $this->getFullRequestURL(), $extWhitelist );
1044                                 if ( $newUrl !== false ) {
1045                                         $this->doSecurityRedirect( $newUrl );
1046                                         return false;
1047                                 }
1048                         }
1049                         throw new HttpError( 403,
1050                                 'Invalid file extension found in the path info or query string.' );
1051                 }
1052                 return true;
1053         }
1054
1055         /**
1056          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1057          * IE 6. Returns true if it was successful, false otherwise.
1058          *
1059          * @param string $url
1060          * @return bool
1061          */
1062         protected function doSecurityRedirect( $url ) {
1063                 header( 'Location: ' . $url );
1064                 header( 'Content-Type: text/html' );
1065                 $encUrl = htmlspecialchars( $url );
1066                 echo <<<HTML
1067 <html>
1068 <head>
1069 <title>Security redirect</title>
1070 </head>
1071 <body>
1072 <h1>Security redirect</h1>
1073 <p>
1074 We can't serve non-HTML content from the URL you have requested, because
1075 Internet Explorer would interpret it as an incorrect and potentially dangerous
1076 content type.</p>
1077 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1078 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1079 Explorer from seeing a bogus file extension.
1080 </p>
1081 </body>
1082 </html>
1083 HTML;
1084                 echo "\n";
1085                 return true;
1086         }
1087
1088         /**
1089          * Parse the Accept-Language header sent by the client into an array
1090          *
1091          * @return array Array( languageCode => q-value ) sorted by q-value in
1092          *   descending order then appearing time in the header in ascending order.
1093          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1094          * This is aligned with rfc2616 section 14.4
1095          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1096          */
1097         public function getAcceptLang() {
1098                 // Modified version of code found at
1099                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1100                 $acceptLang = $this->getHeader( 'Accept-Language' );
1101                 if ( !$acceptLang ) {
1102                         return [];
1103                 }
1104
1105                 // Return the language codes in lower case
1106                 $acceptLang = strtolower( $acceptLang );
1107
1108                 // Break up string into pieces (languages and q factors)
1109                 $lang_parse = null;
1110                 preg_match_all(
1111                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1112                         $acceptLang,
1113                         $lang_parse
1114                 );
1115
1116                 if ( !count( $lang_parse[1] ) ) {
1117                         return [];
1118                 }
1119
1120                 $langcodes = $lang_parse[1];
1121                 $qvalues = $lang_parse[4];
1122                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1123
1124                 // Set default q factor to 1
1125                 foreach ( $indices as $index ) {
1126                         if ( $qvalues[$index] === '' ) {
1127                                 $qvalues[$index] = 1;
1128                         } elseif ( $qvalues[$index] == 0 ) {
1129                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1130                         }
1131                 }
1132
1133                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1134                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1135
1136                 // Create a list like "en" => 0.8
1137                 $langs = array_combine( $langcodes, $qvalues );
1138
1139                 return $langs;
1140         }
1141
1142         /**
1143          * Fetch the raw IP from the request
1144          *
1145          * @since 1.19
1146          *
1147          * @throws MWException
1148          * @return string
1149          */
1150         protected function getRawIP() {
1151                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1152                         return null;
1153                 }
1154
1155                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1156                         throw new MWException( __METHOD__
1157                                 . " : Could not determine the remote IP address due to multiple values." );
1158                 } else {
1159                         $ipchain = $_SERVER['REMOTE_ADDR'];
1160                 }
1161
1162                 return IP::canonicalize( $ipchain );
1163         }
1164
1165         /**
1166          * Work out the IP address based on various globals
1167          * For trusted proxies, use the XFF client IP (first of the chain)
1168          *
1169          * @since 1.19
1170          *
1171          * @throws MWException
1172          * @return string
1173          */
1174         public function getIP() {
1175                 global $wgUsePrivateIPs;
1176
1177                 # Return cached result
1178                 if ( $this->ip !== null ) {
1179                         return $this->ip;
1180                 }
1181
1182                 # collect the originating ips
1183                 $ip = $this->getRawIP();
1184                 if ( !$ip ) {
1185                         throw new MWException( 'Unable to determine IP.' );
1186                 }
1187
1188                 # Append XFF
1189                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1190                 if ( $forwardedFor !== false ) {
1191                         $isConfigured = IP::isConfiguredProxy( $ip );
1192                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1193                         $ipchain = array_reverse( $ipchain );
1194                         array_unshift( $ipchain, $ip );
1195
1196                         # Step through XFF list and find the last address in the list which is a
1197                         # trusted server. Set $ip to the IP address given by that trusted server,
1198                         # unless the address is not sensible (e.g. private). However, prefer private
1199                         # IP addresses over proxy servers controlled by this site (more sensible).
1200                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1201                         foreach ( $ipchain as $i => $curIP ) {
1202                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1203                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1204                                         || !IP::isTrustedProxy( $curIP )
1205                                 ) {
1206                                         break; // IP is not valid/trusted or does not point to anything
1207                                 }
1208                                 if (
1209                                         IP::isPublic( $ipchain[$i + 1] ) ||
1210                                         $wgUsePrivateIPs ||
1211                                         IP::isConfiguredProxy( $curIP ) // bug 48919; treat IP as sane
1212                                 ) {
1213                                         // Follow the next IP according to the proxy
1214                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1215                                         if ( !$nextIP && $isConfigured ) {
1216                                                 // We have not yet made it past CDN/proxy servers of this site,
1217                                                 // so either they are misconfigured or there is some IP spoofing.
1218                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1219                                         }
1220                                         $ip = $nextIP;
1221                                         // keep traversing the chain
1222                                         continue;
1223                                 }
1224                                 break;
1225                         }
1226                 }
1227
1228                 # Allow extensions to improve our guess
1229                 Hooks::run( 'GetIP', [ &$ip ] );
1230
1231                 if ( !$ip ) {
1232                         throw new MWException( "Unable to determine IP." );
1233                 }
1234
1235                 wfDebug( "IP: $ip\n" );
1236                 $this->ip = $ip;
1237                 return $ip;
1238         }
1239
1240         /**
1241          * @param string $ip
1242          * @return void
1243          * @since 1.21
1244          */
1245         public function setIP( $ip ) {
1246                 $this->ip = $ip;
1247         }
1248
1249         /**
1250          * Check if this request uses a "safe" HTTP method
1251          *
1252          * Safe methods are verbs (e.g. GET/HEAD/OPTIONS) used for obtaining content. Such requests
1253          * are not expected to mutate content, especially in ways attributable to the client. Verbs
1254          * like POST and PUT are typical of non-safe requests which often change content.
1255          *
1256          * @return bool
1257          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1258          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1259          * @since 1.28
1260          */
1261         public function hasSafeMethod() {
1262                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1263                         return false; // CLI mode
1264                 }
1265
1266                 return in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS', 'TRACE' ] );
1267         }
1268
1269         /**
1270          * Whether this request should be identified as being "safe"
1271          *
1272          * This means that the client is not requesting any state changes and that database writes
1273          * are not inherently required. Ideally, no visible updates would happen at all. If they
1274          * must, then they should not be publically attributed to the end user.
1275          *
1276          * In more detail:
1277          *   - Cache populations and refreshes MAY occur.
1278          *   - Private user session updates and private server logging MAY occur.
1279          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1280          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1281          *
1282          * @return bool
1283          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1284          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1285          * @since 1.28
1286          */
1287         public function isSafeRequest() {
1288                 if ( $this->markedAsSafe && $this->wasPosted() ) {
1289                         return true; // marked as a "safe" POST
1290                 }
1291
1292                 return $this->hasSafeMethod();
1293         }
1294
1295         /**
1296          * Mark this request as identified as being nullipotent even if it is a POST request
1297          *
1298          * POST requests are often used due to the need for a client payload, even if the request
1299          * is otherwise equivalent to a "safe method" request.
1300          *
1301          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1302          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1303          * @since 1.28
1304          */
1305         public function markAsSafeRequest() {
1306                 $this->markedAsSafe = true;
1307         }
1308 }