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. Used for
 571          * user-supplied freeform text input (for which input transformations may
 572          * be required - e.g.  Esperanto x-coding).
 573          *
 574          * @param string $name
 575          * @param string $default Optional
 576          * @return string
 577          */
 578         public function getText( $name, $default = '' ) {
 579                 global $wgContLang;
 580                 $val = $this->getVal( $name, $default );
 581                 return str_replace( "\r\n", "\n",
 582                         $wgContLang->recodeInput( $val ) );
 583         }
 584
 585         /**
 586          * Extracts the given named values into an array.
 587          * If no arguments are given, returns all input values.
 588          * No transformation is performed on the values.
 589          *
 590          * @return array
 591          */
 592         public function getValues() {
 593                 $names = func_get_args();
 594                 if ( count( $names ) == 0 ) {
 595                         $names = array_keys( $this->data );
 596                 }
 597
 598                 $retVal = [];
 599                 foreach ( $names as $name ) {
 600                         $value = $this->getGPCVal( $this->data, $name, null );
 601                         if ( !is_null( $value ) ) {
 602                                 $retVal[$name] = $value;
 603                         }
 604                 }
 605                 return $retVal;
 606         }
 607
 608         /**
 609          * Returns the names of all input values excluding those in $exclude.
 610          *
 611          * @param array $exclude
 612          * @return array
 613          */
 614         public function getValueNames( $exclude = [] ) {
 615                 return array_diff( array_keys( $this->getValues() ), $exclude );
 616         }
 617
 618         /**
 619          * Get the values passed in the query string.
 620          * No transformation is performed on the values.
 621          *
 622          * @return array
 623          */
 624         public function getQueryValues() {
 625                 return $_GET;
 626         }
 627
 628         /**
 629          * Return the contents of the Query with no decoding. Use when you need to
 630          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 631          *
 632          * @return string
 633          */
 634         public function getRawQueryString() {
 635                 return $_SERVER['QUERY_STRING'];
 636         }
 637
 638         /**
 639          * Return the contents of the POST with no decoding. Use when you need to
 640          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 641          *
 642          * @return string
 643          */
 644         public function getRawPostString() {
 645                 if ( !$this->wasPosted() ) {
 646                         return '';
 647                 }
 648                 return $this->getRawInput();
 649         }
 650
 651         /**
 652          * Return the raw request body, with no processing. Cached since some methods
 653          * disallow reading the stream more than once. As stated in the php docs, this
 654          * does not work with enctype="multipart/form-data".
 655          *
 656          * @return string
 657          */
 658         public function getRawInput() {
 659                 static $input = null;
 660                 if ( $input === null ) {
 661                         $input = file_get_contents( 'php://input' );
 662                 }
 663                 return $input;
 664         }
 665
 666         /**
 667          * Get the HTTP method used for this request.
 668          *
 669          * @return string
 670          */
 671         public function getMethod() {
 672                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 673         }
 674
 675         /**
 676          * Returns true if the present request was reached by a POST operation,
 677          * false otherwise (GET, HEAD, or command-line).
 678          *
 679          * Note that values retrieved by the object may come from the
 680          * GET URL etc even on a POST request.
 681          *
 682          * @return bool
 683          */
 684         public function wasPosted() {
 685                 return $this->getMethod() == 'POST';
 686         }
 687
 688         /**
 689          * Return the session for this request
 690          * @since 1.27
 691          * @note For performance, keep the session locally if you will be making
 692          *  much use of it instead of calling this method repeatedly.
 693          * @return Session
 694          */
 695         public function getSession() {
 696                 if ( $this->sessionId !== null ) {
 697                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 698                         if ( $session ) {
 699                                 return $session;
 700                         }
 701                 }
 702
 703                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 704                 $this->sessionId = $session->getSessionId();
 705                 return $session;
 706         }
 707
 708         /**
 709          * Set the session for this request
 710          * @since 1.27
 711          * @private For use by MediaWiki\Session classes only
 712          * @param SessionId $sessionId
 713          */
 714         public function setSessionId( SessionId $sessionId ) {
 715                 $this->sessionId = $sessionId;
 716         }
 717
 718         /**
 719          * Get the session id for this request, if any
 720          * @since 1.27
 721          * @private For use by MediaWiki\Session classes only
 722          * @return SessionId|null
 723          */
 724         public function getSessionId() {
 725                 return $this->sessionId;
 726         }
 727
 728         /**
 729          * Returns true if the request has a persistent session.
 730          * This does not necessarily mean that the user is logged in!
 731          *
 732          * @deprecated since 1.27, use
 733          *  \MediaWiki\Session\SessionManager::singleton()->getPersistedSessionId()
 734          *  instead.
 735          * @return bool
 736          */
 737         public function checkSessionCookie() {
 738                 global $wgInitialSessionId;
 739                 wfDeprecated( __METHOD__, '1.27' );
 740                 return $wgInitialSessionId !== null &&
 741                         $this->getSession()->getId() === (string)$wgInitialSessionId;
 742         }
 743
 744         /**
 745          * Get a cookie from the $_COOKIE jar
 746          *
 747          * @param string $key The name of the cookie
 748          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 749          * @param mixed $default What to return if the value isn't found
 750          * @return mixed Cookie value or $default if the cookie not set
 751          */
 752         public function getCookie( $key, $prefix = null, $default = null ) {
 753                 if ( $prefix === null ) {
 754                         global $wgCookiePrefix;
 755                         $prefix = $wgCookiePrefix;
 756                 }
 757                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 758         }
 759
 760         /**
 761          * Return the path and query string portion of the main request URI.
 762          * This will be suitable for use as a relative link in HTML output.
 763          *
 764          * @throws MWException
 765          * @return string
 766          */
 767         public static function getGlobalRequestURL() {
 768                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 769                         $base = $_SERVER['REQUEST_URI'];
 770                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 771                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 772                 ) {
 773                         // Probably IIS; doesn't set REQUEST_URI
 774                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 775                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 776                         $base = $_SERVER['SCRIPT_NAME'];
 777                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 778                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 779                         }
 780                 } else {
 781                         // This shouldn't happen!
 782                         throw new MWException( "Web server doesn't provide either " .
 783                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 784                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 785                 }
 786                 // User-agents should not send a fragment with the URI, but
 787                 // if they do, and the web server passes it on to us, we
 788                 // need to strip it or we get false-positive redirect loops
 789                 // or weird output URLs
 790                 $hash = strpos( $base, '#' );
 791                 if ( $hash !== false ) {
 792                         $base = substr( $base, 0, $hash );
 793                 }
 794
 795                 if ( $base[0] == '/' ) {
 796                         // More than one slash will look like it is protocol relative
 797                         return preg_replace( '!^/+!', '/', $base );
 798                 } else {
 799                         // We may get paths with a host prepended; strip it.
 800                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 801                 }
 802         }
 803
 804         /**
 805          * Return the path and query string portion of the request URI.
 806          * This will be suitable for use as a relative link in HTML output.
 807          *
 808          * @throws MWException
 809          * @return string
 810          */
 811         public function getRequestURL() {
 812                 return self::getGlobalRequestURL();
 813         }
 814
 815         /**
 816          * Return the request URI with the canonical service and hostname, path,
 817          * and query string. This will be suitable for use as an absolute link
 818          * in HTML or other output.
 819          *
 820          * If $wgServer is protocol-relative, this will return a fully
 821          * qualified URL with the protocol that was used for this request.
 822          *
 823          * @return string
 824          */
 825         public function getFullRequestURL() {
 826                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 827         }
 828
 829         /**
 830          * @param string $key
 831          * @param string $value
 832          * @return string
 833          */
 834         public function appendQueryValue( $key, $value ) {
 835                 return $this->appendQueryArray( [ $key => $value ] );
 836         }
 837
 838         /**
 839          * Appends or replaces value of query variables.
 840          *
 841          * @param array $array Array of values to replace/add to query
 842          * @return string
 843          */
 844         public function appendQueryArray( $array ) {
 845                 $newquery = $this->getQueryValues();
 846                 unset( $newquery['title'] );
 847                 $newquery = array_merge( $newquery, $array );
 848
 849                 return wfArrayToCgi( $newquery );
 850         }
 851
 852         /**
 853          * Check for limit and offset parameters on the input, and return sensible
 854          * defaults if not given. The limit must be positive and is capped at 5000.
 855          * Offset must be positive but is not capped.
 856          *
 857          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 858          * @param string $optionname To specify an option other than rclimit to pull from.
 859          * @return int[] First element is limit, second is offset
 860          */
 861         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 862                 global $wgUser;
 863
 864                 $limit = $this->getInt( 'limit', 0 );
 865                 if ( $limit < 0 ) {
 866                         $limit = 0;
 867                 }
 868                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 869                         $limit = $wgUser->getIntOption( $optionname );
 870                 }
 871                 if ( $limit <= 0 ) {
 872                         $limit = $deflimit;
 873                 }
 874                 if ( $limit > 5000 ) {
 875                         $limit = 5000; # We have *some* limits...
 876                 }
 877
 878                 $offset = $this->getInt( 'offset', 0 );
 879                 if ( $offset < 0 ) {
 880                         $offset = 0;
 881                 }
 882
 883                 return [ $limit, $offset ];
 884         }
 885
 886         /**
 887          * Return the path to the temporary file where PHP has stored the upload.
 888          *
 889          * @param string $key
 890          * @return string|null String or null if no such file.
 891          */
 892         public function getFileTempname( $key ) {
 893                 $file = new WebRequestUpload( $this, $key );
 894                 return $file->getTempName();
 895         }
 896
 897         /**
 898          * Return the upload error or 0
 899          *
 900          * @param string $key
 901          * @return int
 902          */
 903         public function getUploadError( $key ) {
 904                 $file = new WebRequestUpload( $this, $key );
 905                 return $file->getError();
 906         }
 907
 908         /**
 909          * Return the original filename of the uploaded file, as reported by
 910          * the submitting user agent. HTML-style character entities are
 911          * interpreted and normalized to Unicode normalization form C, in part
 912          * to deal with weird input from Safari with non-ASCII filenames.
 913          *
 914          * Other than this the name is not verified for being a safe filename.
 915          *
 916          * @param string $key
 917          * @return string|null String or null if no such file.
 918          */
 919         public function getFileName( $key ) {
 920                 $file = new WebRequestUpload( $this, $key );
 921                 return $file->getName();
 922         }
 923
 924         /**
 925          * Return a WebRequestUpload object corresponding to the key
 926          *
 927          * @param string $key
 928          * @return WebRequestUpload
 929          */
 930         public function getUpload( $key ) {
 931                 return new WebRequestUpload( $this, $key );
 932         }
 933
 934         /**
 935          * Return a handle to WebResponse style object, for setting cookies,
 936          * headers and other stuff, for Request being worked on.
 937          *
 938          * @return WebResponse
 939          */
 940         public function response() {
 941                 /* Lazy initialization of response object for this request */
 942                 if ( !is_object( $this->response ) ) {
 943                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 944                         $this->response = new $class();
 945                 }
 946                 return $this->response;
 947         }
 948
 949         /**
 950          * Initialise the header list
 951          */
 952         protected function initHeaders() {
 953                 if ( count( $this->headers ) ) {
 954                         return;
 955                 }
 956
 957                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 958                 if ( $apacheHeaders ) {
 959                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 960                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 961                         }
 962                 } else {
 963                         foreach ( $_SERVER as $name => $value ) {
 964                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 965                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 966                                         $this->headers[$name] = $value;
 967                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 968                                         $this->headers['CONTENT-LENGTH'] = $value;
 969                                 }
 970                         }
 971                 }
 972         }
 973
 974         /**
 975          * Get an array containing all request headers
 976          *
 977          * @return array Mapping header name to its value
 978          */
 979         public function getAllHeaders() {
 980                 $this->initHeaders();
 981                 return $this->headers;
 982         }
 983
 984         /**
 985          * Get a request header, or false if it isn't set.
 986          *
 987          * @param string $name Case-insensitive header name
 988          * @param int $flags Bitwise combination of:
 989          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
 990          *                               of values, as described in RFC 2616 § 4.2.
 991          *                               (since 1.26).
 992          * @return string|array|bool False if header is unset; otherwise the
 993          *  header value(s) as either a string (the default) or an array, if
 994          *  WebRequest::GETHEADER_LIST flag was set.
 995          */
 996         public function getHeader( $name, $flags = 0 ) {
 997                 $this->initHeaders();
 998                 $name = strtoupper( $name );
 999                 if ( !isset( $this->headers[$name] ) ) {
1000                         return false;
1001                 }
1002                 $value = $this->headers[$name];
1003                 if ( $flags & self::GETHEADER_LIST ) {
1004                         $value = array_map( 'trim', explode( ',', $value ) );
1005                 }
1006                 return $value;
1007         }
1008
1009         /**
1010          * Get data from the session
1011          *
1012          * @note Prefer $this->getSession() instead if making multiple calls.
1013          * @param string $key Name of key in the session
1014          * @return mixed
1015          */
1016         public function getSessionData( $key ) {
1017                 return $this->getSession()->get( $key );
1018         }
1019
1020         /**
1021          * Set session data
1022          *
1023          * @note Prefer $this->getSession() instead if making multiple calls.
1024          * @param string $key Name of key in the session
1025          * @param mixed $data
1026          */
1027         public function setSessionData( $key, $data ) {
1028                 $this->getSession()->set( $key, $data );
1029         }
1030
1031         /**
1032          * Check if Internet Explorer will detect an incorrect cache extension in
1033          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1034          * message or redirect to a safer URL. Returns true if the URL is OK, and
1035          * false if an error message has been shown and the request should be aborted.
1036          *
1037          * @param array $extWhitelist
1038          * @throws HttpError
1039          * @return bool
1040          */
1041         public function checkUrlExtension( $extWhitelist = [] ) {
1042                 $extWhitelist[] = 'php';
1043                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1044                         if ( !$this->wasPosted() ) {
1045                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1046                                         $this->getFullRequestURL(), $extWhitelist );
1047                                 if ( $newUrl !== false ) {
1048                                         $this->doSecurityRedirect( $newUrl );
1049                                         return false;
1050                                 }
1051                         }
1052                         throw new HttpError( 403,
1053                                 'Invalid file extension found in the path info or query string.' );
1054                 }
1055                 return true;
1056         }
1057
1058         /**
1059          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1060          * IE 6. Returns true if it was successful, false otherwise.
1061          *
1062          * @param string $url
1063          * @return bool
1064          */
1065         protected function doSecurityRedirect( $url ) {
1066                 header( 'Location: ' . $url );
1067                 header( 'Content-Type: text/html' );
1068                 $encUrl = htmlspecialchars( $url );
1069                 echo <<<HTML
1070 <html>
1071 <head>
1072 <title>Security redirect</title>
1073 </head>
1074 <body>
1075 <h1>Security redirect</h1>
1076 <p>
1077 We can't serve non-HTML content from the URL you have requested, because
1078 Internet Explorer would interpret it as an incorrect and potentially dangerous
1079 content type.</p>
1080 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1081 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1082 Explorer from seeing a bogus file extension.
1083 </p>
1084 </body>
1085 </html>
1086 HTML;
1087                 echo "\n";
1088                 return true;
1089         }
1090
1091         /**
1092          * Parse the Accept-Language header sent by the client into an array
1093          *
1094          * @return array Array( languageCode => q-value ) sorted by q-value in
1095          *   descending order then appearing time in the header in ascending order.
1096          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1097          * This is aligned with rfc2616 section 14.4
1098          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1099          */
1100         public function getAcceptLang() {
1101                 // Modified version of code found at
1102                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1103                 $acceptLang = $this->getHeader( 'Accept-Language' );
1104                 if ( !$acceptLang ) {
1105                         return [];
1106                 }
1107
1108                 // Return the language codes in lower case
1109                 $acceptLang = strtolower( $acceptLang );
1110
1111                 // Break up string into pieces (languages and q factors)
1112                 $lang_parse = null;
1113                 preg_match_all(
1114                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1115                         $acceptLang,
1116                         $lang_parse
1117                 );
1118
1119                 if ( !count( $lang_parse[1] ) ) {
1120                         return [];
1121                 }
1122
1123                 $langcodes = $lang_parse[1];
1124                 $qvalues = $lang_parse[4];
1125                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1126
1127                 // Set default q factor to 1
1128                 foreach ( $indices as $index ) {
1129                         if ( $qvalues[$index] === '' ) {
1130                                 $qvalues[$index] = 1;
1131                         } elseif ( $qvalues[$index] == 0 ) {
1132                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1133                         }
1134                 }
1135
1136                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1137                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1138
1139                 // Create a list like "en" => 0.8
1140                 $langs = array_combine( $langcodes, $qvalues );
1141
1142                 return $langs;
1143         }
1144
1145         /**
1146          * Fetch the raw IP from the request
1147          *
1148          * @since 1.19
1149          *
1150          * @throws MWException
1151          * @return string
1152          */
1153         protected function getRawIP() {
1154                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1155                         return null;
1156                 }
1157
1158                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1159                         throw new MWException( __METHOD__
1160                                 . " : Could not determine the remote IP address due to multiple values." );
1161                 } else {
1162                         $ipchain = $_SERVER['REMOTE_ADDR'];
1163                 }
1164
1165                 return IP::canonicalize( $ipchain );
1166         }
1167
1168         /**
1169          * Work out the IP address based on various globals
1170          * For trusted proxies, use the XFF client IP (first of the chain)
1171          *
1172          * @since 1.19
1173          *
1174          * @throws MWException
1175          * @return string
1176          */
1177         public function getIP() {
1178                 global $wgUsePrivateIPs;
1179
1180                 # Return cached result
1181                 if ( $this->ip !== null ) {
1182                         return $this->ip;
1183                 }
1184
1185                 # collect the originating ips
1186                 $ip = $this->getRawIP();
1187                 if ( !$ip ) {
1188                         throw new MWException( 'Unable to determine IP.' );
1189                 }
1190
1191                 # Append XFF
1192                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1193                 if ( $forwardedFor !== false ) {
1194                         $isConfigured = IP::isConfiguredProxy( $ip );
1195                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1196                         $ipchain = array_reverse( $ipchain );
1197                         array_unshift( $ipchain, $ip );
1198
1199                         # Step through XFF list and find the last address in the list which is a
1200                         # trusted server. Set $ip to the IP address given by that trusted server,
1201                         # unless the address is not sensible (e.g. private). However, prefer private
1202                         # IP addresses over proxy servers controlled by this site (more sensible).
1203                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1204                         foreach ( $ipchain as $i => $curIP ) {
1205                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1206                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1207                                         || !IP::isTrustedProxy( $curIP )
1208                                 ) {
1209                                         break; // IP is not valid/trusted or does not point to anything
1210                                 }
1211                                 if (
1212                                         IP::isPublic( $ipchain[$i + 1] ) ||
1213                                         $wgUsePrivateIPs ||
1214                                         IP::isConfiguredProxy( $curIP ) // bug 48919; treat IP as sane
1215                                 ) {
1216                                         // Follow the next IP according to the proxy
1217                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1218                                         if ( !$nextIP && $isConfigured ) {
1219                                                 // We have not yet made it past CDN/proxy servers of this site,
1220                                                 // so either they are misconfigured or there is some IP spoofing.
1221                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1222                                         }
1223                                         $ip = $nextIP;
1224                                         // keep traversing the chain
1225                                         continue;
1226                                 }
1227                                 break;
1228                         }
1229                 }
1230
1231                 # Allow extensions to improve our guess
1232                 Hooks::run( 'GetIP', [ &$ip ] );
1233
1234                 if ( !$ip ) {
1235                         throw new MWException( "Unable to determine IP." );
1236                 }
1237
1238                 wfDebug( "IP: $ip\n" );
1239                 $this->ip = $ip;
1240                 return $ip;
1241         }
1242
1243         /**
1244          * @param string $ip
1245          * @return void
1246          * @since 1.21
1247          */
1248         public function setIP( $ip ) {
1249                 $this->ip = $ip;
1250         }
1251
1252         /**
1253          * Whether this request should be identified as being "safe"
1254          *
1255          * This means that the client is not requesting any state changes and that database writes
1256          * are not inherently required. Ideally, no visible updates would happen at all. If they
1257          * must, then they should not be publically attributed to the end user.
1258          *
1259          * In more detail:
1260          *   - Cache populations and refreshes MAY occur.
1261          *   - Private user session updates and private server logging MAY occur.
1262          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1263          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1264          *
1265          * @return bool
1266          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1267          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1268          * @since 1.28
1269          */
1270         public function isSafeRequest() {
1271                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1272                         return false; // CLI mode
1273                 }
1274
1275                 if ( $_SERVER['REQUEST_METHOD'] === 'POST' ) {
1276                         return $this->markedAsSafe;
1277                 } elseif ( in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS' ] ) ) {
1278                         return true; // HTTP "safe methods"
1279                 }
1280
1281                 return false; // PUT/DELETE
1282         }
1283
1284         /**
1285          * Mark this request is identified as being nullipotent even if it is a POST request
1286          *
1287          * POST requests are often used due to the need for a client payload, even if the request
1288          * is otherwise equivalent to a "safe method" request.
1289          *
1290          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1291          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1292          * @since 1.28
1293          */
1294         public function markAsSafeRequest() {
1295                 $this->markedAsSafe = true;
1296         }
1297 }