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 without normalization, or return $default
 399          * if it's not set.
 400          *
 401          * Unlike self::getVal(), this does not perform any normalization on the
 402          * input value.
 403          *
 404          * @since 1.28
 405          * @param string $name
 406          * @param string|null $default Optional default
 407          * @return string
 408          */
 409         public function getRawVal( $name, $default = null ) {
 410                 $name = strtr( $name, '.', '_' ); // See comment in self::getGPCVal()
 411                 if ( isset( $this->data[$name] ) && !is_array( $this->data[$name] ) ) {
 412                         $val = $this->data[$name];
 413                 } else {
 414                         $val = $default;
 415                 }
 416                 if ( is_null( $val ) ) {
 417                         return $val;
 418                 } else {
 419                         return (string)$val;
 420                 }
 421         }
 422
 423         /**
 424          * Fetch a scalar from the input or return $default if it's not set.
 425          * Returns a string. Arrays are discarded. Useful for
 426          * non-freeform text inputs (e.g. predefined internal text keys
 427          * selected by a drop-down menu). For freeform input, see getText().
 428          *
 429          * @param string $name
 430          * @param string $default Optional default (or null)
 431          * @return string
 432          */
 433         public function getVal( $name, $default = null ) {
 434                 $val = $this->getGPCVal( $this->data, $name, $default );
 435                 if ( is_array( $val ) ) {
 436                         $val = $default;
 437                 }
 438                 if ( is_null( $val ) ) {
 439                         return $val;
 440                 } else {
 441                         return (string)$val;
 442                 }
 443         }
 444
 445         /**
 446          * Set an arbitrary value into our get/post data.
 447          *
 448          * @param string $key Key name to use
 449          * @param mixed $value Value to set
 450          * @return mixed Old value if one was present, null otherwise
 451          */
 452         public function setVal( $key, $value ) {
 453                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 454                 $this->data[$key] = $value;
 455                 return $ret;
 456         }
 457
 458         /**
 459          * Unset an arbitrary value from our get/post data.
 460          *
 461          * @param string $key Key name to use
 462          * @return mixed Old value if one was present, null otherwise
 463          */
 464         public function unsetVal( $key ) {
 465                 if ( !isset( $this->data[$key] ) ) {
 466                         $ret = null;
 467                 } else {
 468                         $ret = $this->data[$key];
 469                         unset( $this->data[$key] );
 470                 }
 471                 return $ret;
 472         }
 473
 474         /**
 475          * Fetch an array from the input or return $default if it's not set.
 476          * If source was scalar, will return an array with a single element.
 477          * If no source and no default, returns null.
 478          *
 479          * @param string $name
 480          * @param array $default Optional default (or null)
 481          * @return array
 482          */
 483         public function getArray( $name, $default = null ) {
 484                 $val = $this->getGPCVal( $this->data, $name, $default );
 485                 if ( is_null( $val ) ) {
 486                         return null;
 487                 } else {
 488                         return (array)$val;
 489                 }
 490         }
 491
 492         /**
 493          * Fetch an array of integers, or return $default if it's not set.
 494          * If source was scalar, will return an array with a single element.
 495          * If no source and no default, returns null.
 496          * If an array is returned, contents are guaranteed to be integers.
 497          *
 498          * @param string $name
 499          * @param array $default Option default (or null)
 500          * @return array Array of ints
 501          */
 502         public function getIntArray( $name, $default = null ) {
 503                 $val = $this->getArray( $name, $default );
 504                 if ( is_array( $val ) ) {
 505                         $val = array_map( 'intval', $val );
 506                 }
 507                 return $val;
 508         }
 509
 510         /**
 511          * Fetch an integer value from the input or return $default if not set.
 512          * Guaranteed to return an integer; non-numeric input will typically
 513          * return 0.
 514          *
 515          * @param string $name
 516          * @param int $default
 517          * @return int
 518          */
 519         public function getInt( $name, $default = 0 ) {
 520                 return intval( $this->getVal( $name, $default ) );
 521         }
 522
 523         /**
 524          * Fetch an integer value from the input or return null if empty.
 525          * Guaranteed to return an integer or null; non-numeric input will
 526          * typically return null.
 527          *
 528          * @param string $name
 529          * @return int|null
 530          */
 531         public function getIntOrNull( $name ) {
 532                 $val = $this->getVal( $name );
 533                 return is_numeric( $val )
 534                         ? intval( $val )
 535                         : null;
 536         }
 537
 538         /**
 539          * Fetch a floating point value from the input or return $default if not set.
 540          * Guaranteed to return a float; non-numeric input will typically
 541          * return 0.
 542          *
 543          * @since 1.23
 544          * @param string $name
 545          * @param float $default
 546          * @return float
 547          */
 548         public function getFloat( $name, $default = 0.0 ) {
 549                 return floatval( $this->getVal( $name, $default ) );
 550         }
 551
 552         /**
 553          * Fetch a boolean value from the input or return $default if not set.
 554          * Guaranteed to return true or false, with normal PHP semantics for
 555          * boolean interpretation of strings.
 556          *
 557          * @param string $name
 558          * @param bool $default
 559          * @return bool
 560          */
 561         public function getBool( $name, $default = false ) {
 562                 return (bool)$this->getVal( $name, $default );
 563         }
 564
 565         /**
 566          * Fetch a boolean value from the input or return $default if not set.
 567          * Unlike getBool, the string "false" will result in boolean false, which is
 568          * useful when interpreting information sent from JavaScript.
 569          *
 570          * @param string $name
 571          * @param bool $default
 572          * @return bool
 573          */
 574         public function getFuzzyBool( $name, $default = false ) {
 575                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 576         }
 577
 578         /**
 579          * Return true if the named value is set in the input, whatever that
 580          * value is (even "0"). Return false if the named value is not set.
 581          * Example use is checking for the presence of check boxes in forms.
 582          *
 583          * @param string $name
 584          * @return bool
 585          */
 586         public function getCheck( $name ) {
 587                 # Checkboxes and buttons are only present when clicked
 588                 # Presence connotes truth, absence false
 589                 return $this->getVal( $name, null ) !== null;
 590         }
 591
 592         /**
 593          * Fetch a text string from the given array or return $default if it's not
 594          * set. Carriage returns are stripped from the text. This should generally
 595          * be used for form "<textarea>" and "<input>" fields, and for
 596          * user-supplied freeform text input.
 597          *
 598          * @param string $name
 599          * @param string $default Optional
 600          * @return string
 601          */
 602         public function getText( $name, $default = '' ) {
 603                 $val = $this->getVal( $name, $default );
 604                 return str_replace( "\r\n", "\n", $val );
 605         }
 606
 607         /**
 608          * Extracts the given named values into an array.
 609          * If no arguments are given, returns all input values.
 610          * No transformation is performed on the values.
 611          *
 612          * @return array
 613          */
 614         public function getValues() {
 615                 $names = func_get_args();
 616                 if ( count( $names ) == 0 ) {
 617                         $names = array_keys( $this->data );
 618                 }
 619
 620                 $retVal = [];
 621                 foreach ( $names as $name ) {
 622                         $value = $this->getGPCVal( $this->data, $name, null );
 623                         if ( !is_null( $value ) ) {
 624                                 $retVal[$name] = $value;
 625                         }
 626                 }
 627                 return $retVal;
 628         }
 629
 630         /**
 631          * Returns the names of all input values excluding those in $exclude.
 632          *
 633          * @param array $exclude
 634          * @return array
 635          */
 636         public function getValueNames( $exclude = [] ) {
 637                 return array_diff( array_keys( $this->getValues() ), $exclude );
 638         }
 639
 640         /**
 641          * Get the values passed in the query string.
 642          * No transformation is performed on the values.
 643          *
 644          * @return array
 645          */
 646         public function getQueryValues() {
 647                 return $_GET;
 648         }
 649
 650         /**
 651          * Return the contents of the Query with no decoding. Use when you need to
 652          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 653          *
 654          * @return string
 655          */
 656         public function getRawQueryString() {
 657                 return $_SERVER['QUERY_STRING'];
 658         }
 659
 660         /**
 661          * Return the contents of the POST with no decoding. Use when you need to
 662          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 663          *
 664          * @return string
 665          */
 666         public function getRawPostString() {
 667                 if ( !$this->wasPosted() ) {
 668                         return '';
 669                 }
 670                 return $this->getRawInput();
 671         }
 672
 673         /**
 674          * Return the raw request body, with no processing. Cached since some methods
 675          * disallow reading the stream more than once. As stated in the php docs, this
 676          * does not work with enctype="multipart/form-data".
 677          *
 678          * @return string
 679          */
 680         public function getRawInput() {
 681                 static $input = null;
 682                 if ( $input === null ) {
 683                         $input = file_get_contents( 'php://input' );
 684                 }
 685                 return $input;
 686         }
 687
 688         /**
 689          * Get the HTTP method used for this request.
 690          *
 691          * @return string
 692          */
 693         public function getMethod() {
 694                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 695         }
 696
 697         /**
 698          * Returns true if the present request was reached by a POST operation,
 699          * false otherwise (GET, HEAD, or command-line).
 700          *
 701          * Note that values retrieved by the object may come from the
 702          * GET URL etc even on a POST request.
 703          *
 704          * @return bool
 705          */
 706         public function wasPosted() {
 707                 return $this->getMethod() == 'POST';
 708         }
 709
 710         /**
 711          * Return the session for this request
 712          *
 713          * This might unpersist an existing session if it was invalid.
 714          *
 715          * @since 1.27
 716          * @note For performance, keep the session locally if you will be making
 717          *  much use of it instead of calling this method repeatedly.
 718          * @return Session
 719          */
 720         public function getSession() {
 721                 if ( $this->sessionId !== null ) {
 722                         $session = SessionManager::singleton()->getSessionById( (string)$this->sessionId, true, $this );
 723                         if ( $session ) {
 724                                 return $session;
 725                         }
 726                 }
 727
 728                 $session = SessionManager::singleton()->getSessionForRequest( $this );
 729                 $this->sessionId = $session->getSessionId();
 730                 return $session;
 731         }
 732
 733         /**
 734          * Set the session for this request
 735          * @since 1.27
 736          * @private For use by MediaWiki\Session classes only
 737          * @param SessionId $sessionId
 738          */
 739         public function setSessionId( SessionId $sessionId ) {
 740                 $this->sessionId = $sessionId;
 741         }
 742
 743         /**
 744          * Get the session id for this request, if any
 745          * @since 1.27
 746          * @private For use by MediaWiki\Session classes only
 747          * @return SessionId|null
 748          */
 749         public function getSessionId() {
 750                 return $this->sessionId;
 751         }
 752
 753         /**
 754          * Returns true if the request has a persistent session.
 755          * This does not necessarily mean that the user is logged in!
 756          *
 757          * @deprecated since 1.27, use
 758          *  \MediaWiki\Session\SessionManager::singleton()->getPersistedSessionId()
 759          *  instead.
 760          * @return bool
 761          */
 762         public function checkSessionCookie() {
 763                 global $wgInitialSessionId;
 764                 wfDeprecated( __METHOD__, '1.27' );
 765                 return $wgInitialSessionId !== null &&
 766                         $this->getSession()->getId() === (string)$wgInitialSessionId;
 767         }
 768
 769         /**
 770          * Get a cookie from the $_COOKIE jar
 771          *
 772          * @param string $key The name of the cookie
 773          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 774          * @param mixed $default What to return if the value isn't found
 775          * @return mixed Cookie value or $default if the cookie not set
 776          */
 777         public function getCookie( $key, $prefix = null, $default = null ) {
 778                 if ( $prefix === null ) {
 779                         global $wgCookiePrefix;
 780                         $prefix = $wgCookiePrefix;
 781                 }
 782                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 783         }
 784
 785         /**
 786          * Return the path and query string portion of the main request URI.
 787          * This will be suitable for use as a relative link in HTML output.
 788          *
 789          * @throws MWException
 790          * @return string
 791          */
 792         public static function getGlobalRequestURL() {
 793                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 794                         $base = $_SERVER['REQUEST_URI'];
 795                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 796                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 797                 ) {
 798                         // Probably IIS; doesn't set REQUEST_URI
 799                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 800                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 801                         $base = $_SERVER['SCRIPT_NAME'];
 802                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 803                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 804                         }
 805                 } else {
 806                         // This shouldn't happen!
 807                         throw new MWException( "Web server doesn't provide either " .
 808                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 809                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 810                 }
 811                 // User-agents should not send a fragment with the URI, but
 812                 // if they do, and the web server passes it on to us, we
 813                 // need to strip it or we get false-positive redirect loops
 814                 // or weird output URLs
 815                 $hash = strpos( $base, '#' );
 816                 if ( $hash !== false ) {
 817                         $base = substr( $base, 0, $hash );
 818                 }
 819
 820                 if ( $base[0] == '/' ) {
 821                         // More than one slash will look like it is protocol relative
 822                         return preg_replace( '!^/+!', '/', $base );
 823                 } else {
 824                         // We may get paths with a host prepended; strip it.
 825                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 826                 }
 827         }
 828
 829         /**
 830          * Return the path and query string portion of the request URI.
 831          * This will be suitable for use as a relative link in HTML output.
 832          *
 833          * @throws MWException
 834          * @return string
 835          */
 836         public function getRequestURL() {
 837                 return self::getGlobalRequestURL();
 838         }
 839
 840         /**
 841          * Return the request URI with the canonical service and hostname, path,
 842          * and query string. This will be suitable for use as an absolute link
 843          * in HTML or other output.
 844          *
 845          * If $wgServer is protocol-relative, this will return a fully
 846          * qualified URL with the protocol that was used for this request.
 847          *
 848          * @return string
 849          */
 850         public function getFullRequestURL() {
 851                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 852         }
 853
 854         /**
 855          * @param string $key
 856          * @param string $value
 857          * @return string
 858          */
 859         public function appendQueryValue( $key, $value ) {
 860                 return $this->appendQueryArray( [ $key => $value ] );
 861         }
 862
 863         /**
 864          * Appends or replaces value of query variables.
 865          *
 866          * @param array $array Array of values to replace/add to query
 867          * @return string
 868          */
 869         public function appendQueryArray( $array ) {
 870                 $newquery = $this->getQueryValues();
 871                 unset( $newquery['title'] );
 872                 $newquery = array_merge( $newquery, $array );
 873
 874                 return wfArrayToCgi( $newquery );
 875         }
 876
 877         /**
 878          * Check for limit and offset parameters on the input, and return sensible
 879          * defaults if not given. The limit must be positive and is capped at 5000.
 880          * Offset must be positive but is not capped.
 881          *
 882          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 883          * @param string $optionname To specify an option other than rclimit to pull from.
 884          * @return int[] First element is limit, second is offset
 885          */
 886         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 887                 global $wgUser;
 888
 889                 $limit = $this->getInt( 'limit', 0 );
 890                 if ( $limit < 0 ) {
 891                         $limit = 0;
 892                 }
 893                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 894                         $limit = $wgUser->getIntOption( $optionname );
 895                 }
 896                 if ( $limit <= 0 ) {
 897                         $limit = $deflimit;
 898                 }
 899                 if ( $limit > 5000 ) {
 900                         $limit = 5000; # We have *some* limits...
 901                 }
 902
 903                 $offset = $this->getInt( 'offset', 0 );
 904                 if ( $offset < 0 ) {
 905                         $offset = 0;
 906                 }
 907
 908                 return [ $limit, $offset ];
 909         }
 910
 911         /**
 912          * Return the path to the temporary file where PHP has stored the upload.
 913          *
 914          * @param string $key
 915          * @return string|null String or null if no such file.
 916          */
 917         public function getFileTempname( $key ) {
 918                 $file = new WebRequestUpload( $this, $key );
 919                 return $file->getTempName();
 920         }
 921
 922         /**
 923          * Return the upload error or 0
 924          *
 925          * @param string $key
 926          * @return int
 927          */
 928         public function getUploadError( $key ) {
 929                 $file = new WebRequestUpload( $this, $key );
 930                 return $file->getError();
 931         }
 932
 933         /**
 934          * Return the original filename of the uploaded file, as reported by
 935          * the submitting user agent. HTML-style character entities are
 936          * interpreted and normalized to Unicode normalization form C, in part
 937          * to deal with weird input from Safari with non-ASCII filenames.
 938          *
 939          * Other than this the name is not verified for being a safe filename.
 940          *
 941          * @param string $key
 942          * @return string|null String or null if no such file.
 943          */
 944         public function getFileName( $key ) {
 945                 $file = new WebRequestUpload( $this, $key );
 946                 return $file->getName();
 947         }
 948
 949         /**
 950          * Return a WebRequestUpload object corresponding to the key
 951          *
 952          * @param string $key
 953          * @return WebRequestUpload
 954          */
 955         public function getUpload( $key ) {
 956                 return new WebRequestUpload( $this, $key );
 957         }
 958
 959         /**
 960          * Return a handle to WebResponse style object, for setting cookies,
 961          * headers and other stuff, for Request being worked on.
 962          *
 963          * @return WebResponse
 964          */
 965         public function response() {
 966                 /* Lazy initialization of response object for this request */
 967                 if ( !is_object( $this->response ) ) {
 968                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 969                         $this->response = new $class();
 970                 }
 971                 return $this->response;
 972         }
 973
 974         /**
 975          * Initialise the header list
 976          */
 977         protected function initHeaders() {
 978                 if ( count( $this->headers ) ) {
 979                         return;
 980                 }
 981
 982                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 983                 if ( $apacheHeaders ) {
 984                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 985                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 986                         }
 987                 } else {
 988                         foreach ( $_SERVER as $name => $value ) {
 989                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 990                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 991                                         $this->headers[$name] = $value;
 992                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 993                                         $this->headers['CONTENT-LENGTH'] = $value;
 994                                 }
 995                         }
 996                 }
 997         }
 998
 999         /**
1000          * Get an array containing all request headers
1001          *
1002          * @return array Mapping header name to its value
1003          */
1004         public function getAllHeaders() {
1005                 $this->initHeaders();
1006                 return $this->headers;
1007         }
1008
1009         /**
1010          * Get a request header, or false if it isn't set.
1011          *
1012          * @param string $name Case-insensitive header name
1013          * @param int $flags Bitwise combination of:
1014          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
1015          *                               of values, as described in RFC 2616 § 4.2.
1016          *                               (since 1.26).
1017          * @return string|array|bool False if header is unset; otherwise the
1018          *  header value(s) as either a string (the default) or an array, if
1019          *  WebRequest::GETHEADER_LIST flag was set.
1020          */
1021         public function getHeader( $name, $flags = 0 ) {
1022                 $this->initHeaders();
1023                 $name = strtoupper( $name );
1024                 if ( !isset( $this->headers[$name] ) ) {
1025                         return false;
1026                 }
1027                 $value = $this->headers[$name];
1028                 if ( $flags & self::GETHEADER_LIST ) {
1029                         $value = array_map( 'trim', explode( ',', $value ) );
1030                 }
1031                 return $value;
1032         }
1033
1034         /**
1035          * Get data from the session
1036          *
1037          * @note Prefer $this->getSession() instead if making multiple calls.
1038          * @param string $key Name of key in the session
1039          * @return mixed
1040          */
1041         public function getSessionData( $key ) {
1042                 return $this->getSession()->get( $key );
1043         }
1044
1045         /**
1046          * Set session data
1047          *
1048          * @note Prefer $this->getSession() instead if making multiple calls.
1049          * @param string $key Name of key in the session
1050          * @param mixed $data
1051          */
1052         public function setSessionData( $key, $data ) {
1053                 $this->getSession()->set( $key, $data );
1054         }
1055
1056         /**
1057          * Check if Internet Explorer will detect an incorrect cache extension in
1058          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
1059          * message or redirect to a safer URL. Returns true if the URL is OK, and
1060          * false if an error message has been shown and the request should be aborted.
1061          *
1062          * @param array $extWhitelist
1063          * @throws HttpError
1064          * @return bool
1065          */
1066         public function checkUrlExtension( $extWhitelist = [] ) {
1067                 $extWhitelist[] = 'php';
1068                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
1069                         if ( !$this->wasPosted() ) {
1070                                 $newUrl = IEUrlExtension::fixUrlForIE6(
1071                                         $this->getFullRequestURL(), $extWhitelist );
1072                                 if ( $newUrl !== false ) {
1073                                         $this->doSecurityRedirect( $newUrl );
1074                                         return false;
1075                                 }
1076                         }
1077                         throw new HttpError( 403,
1078                                 'Invalid file extension found in the path info or query string.' );
1079                 }
1080                 return true;
1081         }
1082
1083         /**
1084          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
1085          * IE 6. Returns true if it was successful, false otherwise.
1086          *
1087          * @param string $url
1088          * @return bool
1089          */
1090         protected function doSecurityRedirect( $url ) {
1091                 header( 'Location: ' . $url );
1092                 header( 'Content-Type: text/html' );
1093                 $encUrl = htmlspecialchars( $url );
1094                 echo <<<HTML
1095 <html>
1096 <head>
1097 <title>Security redirect</title>
1098 </head>
1099 <body>
1100 <h1>Security redirect</h1>
1101 <p>
1102 We can't serve non-HTML content from the URL you have requested, because
1103 Internet Explorer would interpret it as an incorrect and potentially dangerous
1104 content type.</p>
1105 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
1106 URL you have requested, except that "&amp;*" is appended. This prevents Internet
1107 Explorer from seeing a bogus file extension.
1108 </p>
1109 </body>
1110 </html>
1111 HTML;
1112                 echo "\n";
1113                 return true;
1114         }
1115
1116         /**
1117          * Parse the Accept-Language header sent by the client into an array
1118          *
1119          * @return array Array( languageCode => q-value ) sorted by q-value in
1120          *   descending order then appearing time in the header in ascending order.
1121          * May contain the "language" '*', which applies to languages other than those explicitly listed.
1122          * This is aligned with rfc2616 section 14.4
1123          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1124          */
1125         public function getAcceptLang() {
1126                 // Modified version of code found at
1127                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1128                 $acceptLang = $this->getHeader( 'Accept-Language' );
1129                 if ( !$acceptLang ) {
1130                         return [];
1131                 }
1132
1133                 // Return the language codes in lower case
1134                 $acceptLang = strtolower( $acceptLang );
1135
1136                 // Break up string into pieces (languages and q factors)
1137                 $lang_parse = null;
1138                 preg_match_all(
1139                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1140                         $acceptLang,
1141                         $lang_parse
1142                 );
1143
1144                 if ( !count( $lang_parse[1] ) ) {
1145                         return [];
1146                 }
1147
1148                 $langcodes = $lang_parse[1];
1149                 $qvalues = $lang_parse[4];
1150                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1151
1152                 // Set default q factor to 1
1153                 foreach ( $indices as $index ) {
1154                         if ( $qvalues[$index] === '' ) {
1155                                 $qvalues[$index] = 1;
1156                         } elseif ( $qvalues[$index] == 0 ) {
1157                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1158                         }
1159                 }
1160
1161                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1162                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1163
1164                 // Create a list like "en" => 0.8
1165                 $langs = array_combine( $langcodes, $qvalues );
1166
1167                 return $langs;
1168         }
1169
1170         /**
1171          * Fetch the raw IP from the request
1172          *
1173          * @since 1.19
1174          *
1175          * @throws MWException
1176          * @return string
1177          */
1178         protected function getRawIP() {
1179                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1180                         return null;
1181                 }
1182
1183                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1184                         throw new MWException( __METHOD__
1185                                 . " : Could not determine the remote IP address due to multiple values." );
1186                 } else {
1187                         $ipchain = $_SERVER['REMOTE_ADDR'];
1188                 }
1189
1190                 return IP::canonicalize( $ipchain );
1191         }
1192
1193         /**
1194          * Work out the IP address based on various globals
1195          * For trusted proxies, use the XFF client IP (first of the chain)
1196          *
1197          * @since 1.19
1198          *
1199          * @throws MWException
1200          * @return string
1201          */
1202         public function getIP() {
1203                 global $wgUsePrivateIPs;
1204
1205                 # Return cached result
1206                 if ( $this->ip !== null ) {
1207                         return $this->ip;
1208                 }
1209
1210                 # collect the originating ips
1211                 $ip = $this->getRawIP();
1212                 if ( !$ip ) {
1213                         throw new MWException( 'Unable to determine IP.' );
1214                 }
1215
1216                 # Append XFF
1217                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1218                 if ( $forwardedFor !== false ) {
1219                         $isConfigured = IP::isConfiguredProxy( $ip );
1220                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1221                         $ipchain = array_reverse( $ipchain );
1222                         array_unshift( $ipchain, $ip );
1223
1224                         # Step through XFF list and find the last address in the list which is a
1225                         # trusted server. Set $ip to the IP address given by that trusted server,
1226                         # unless the address is not sensible (e.g. private). However, prefer private
1227                         # IP addresses over proxy servers controlled by this site (more sensible).
1228                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1229                         foreach ( $ipchain as $i => $curIP ) {
1230                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1231                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1232                                         || !IP::isTrustedProxy( $curIP )
1233                                 ) {
1234                                         break; // IP is not valid/trusted or does not point to anything
1235                                 }
1236                                 if (
1237                                         IP::isPublic( $ipchain[$i + 1] ) ||
1238                                         $wgUsePrivateIPs ||
1239                                         IP::isConfiguredProxy( $curIP ) // bug 48919; treat IP as sane
1240                                 ) {
1241                                         // Follow the next IP according to the proxy
1242                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1243                                         if ( !$nextIP && $isConfigured ) {
1244                                                 // We have not yet made it past CDN/proxy servers of this site,
1245                                                 // so either they are misconfigured or there is some IP spoofing.
1246                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1247                                         }
1248                                         $ip = $nextIP;
1249                                         // keep traversing the chain
1250                                         continue;
1251                                 }
1252                                 break;
1253                         }
1254                 }
1255
1256                 # Allow extensions to improve our guess
1257                 Hooks::run( 'GetIP', [ &$ip ] );
1258
1259                 if ( !$ip ) {
1260                         throw new MWException( "Unable to determine IP." );
1261                 }
1262
1263                 wfDebug( "IP: $ip\n" );
1264                 $this->ip = $ip;
1265                 return $ip;
1266         }
1267
1268         /**
1269          * @param string $ip
1270          * @return void
1271          * @since 1.21
1272          */
1273         public function setIP( $ip ) {
1274                 $this->ip = $ip;
1275         }
1276
1277         /**
1278          * Check if this request uses a "safe" HTTP method
1279          *
1280          * Safe methods are verbs (e.g. GET/HEAD/OPTIONS) used for obtaining content. Such requests
1281          * are not expected to mutate content, especially in ways attributable to the client. Verbs
1282          * like POST and PUT are typical of non-safe requests which often change content.
1283          *
1284          * @return bool
1285          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1286          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1287          * @since 1.28
1288          */
1289         public function hasSafeMethod() {
1290                 if ( !isset( $_SERVER['REQUEST_METHOD'] ) ) {
1291                         return false; // CLI mode
1292                 }
1293
1294                 return in_array( $_SERVER['REQUEST_METHOD'], [ 'GET', 'HEAD', 'OPTIONS', 'TRACE' ] );
1295         }
1296
1297         /**
1298          * Whether this request should be identified as being "safe"
1299          *
1300          * This means that the client is not requesting any state changes and that database writes
1301          * are not inherently required. Ideally, no visible updates would happen at all. If they
1302          * must, then they should not be publically attributed to the end user.
1303          *
1304          * In more detail:
1305          *   - Cache populations and refreshes MAY occur.
1306          *   - Private user session updates and private server logging MAY occur.
1307          *   - Updates to private viewing activity data MAY occur via DeferredUpdates.
1308          *   - Other updates SHOULD NOT occur (e.g. modifying content assets).
1309          *
1310          * @return bool
1311          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1312          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1313          * @since 1.28
1314          */
1315         public function isSafeRequest() {
1316                 if ( $this->markedAsSafe && $this->wasPosted() ) {
1317                         return true; // marked as a "safe" POST
1318                 }
1319
1320                 return $this->hasSafeMethod();
1321         }
1322
1323         /**
1324          * Mark this request as identified as being nullipotent even if it is a POST request
1325          *
1326          * POST requests are often used due to the need for a client payload, even if the request
1327          * is otherwise equivalent to a "safe method" request.
1328          *
1329          * @see https://tools.ietf.org/html/rfc7231#section-4.2.1
1330          * @see https://www.w3.org/Protocols/rfc2616/rfc2616-sec9.html
1331          * @since 1.28
1332          */
1333         public function markAsSafeRequest() {
1334                 $this->markedAsSafe = true;
1335         }
1336 }