includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasssty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * http://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 /**
  27  * The WebRequest class encapsulates getting at data passed in the
  28  * URL or via a POSTed form, handling remove of "magic quotes" slashes,
  29  * stripping illegal input characters and normalizing Unicode sequences.
  30  *
  31  * Usually this is used via a global singleton, $wgRequest. You should
  32  * not create a second WebRequest object; make a FauxRequest object if
  33  * you want to pass arbitrary data to some function in place of the web
  34  * input.
  35  *
  36  * @ingroup HTTP
  37  */
  38 class WebRequest {
  39         protected $data, $headers = array();
  40
  41         /**
  42          * Lazy-init response object
  43          * @var WebResponse
  44          */
  45         private $response;
  46
  47         /**
  48          * Cached client IP address
  49          * @var String
  50          */
  51         private $ip;
  52
  53         public function __construct() {
  54                 /// @todo FIXME: This preemptive de-quoting can interfere with other web libraries
  55                 ///        and increases our memory footprint. It would be cleaner to do on
  56                 ///        demand; but currently we have no wrapper for $_SERVER etc.
  57                 $this->checkMagicQuotes();
  58
  59                 // POST overrides GET data
  60                 // We don't use $_REQUEST here to avoid interference from cookies...
  61                 $this->data = $_POST + $_GET;
  62         }
  63
  64         /**
  65          * Extract relevant query arguments from the http request uri's path
  66          * to be merged with the normal php provided query arguments.
  67          * Tries to use the REQUEST_URI data if available and parses it
  68          * according to the wiki's configuration looking for any known pattern.
  69          *
  70          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
  71          * provided by the server if any and use that to set a 'title' parameter.
  72          *
  73          * @param $want string: If this is not 'all', then the function
  74          * will return an empty array if it determines that the URL is
  75          * inside a rewrite path.
  76          *
  77          * @return Array: Any query arguments found in path matches.
  78          */
  79         static public function getPathInfo( $want = 'all' ) {
  80                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
  81                 // And also by Apache 2.x, double slashes are converted to single slashes.
  82                 // So we will use REQUEST_URI if possible.
  83                 $matches = array();
  84                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
  85                         // Slurp out the path portion to examine...
  86                         $url = $_SERVER['REQUEST_URI'];
  87                         if ( !preg_match( '!^https?://!', $url ) ) {
  88                                 $url = 'http://unused' . $url;
  89                         }
  90                         $a = parse_url( $url );
  91                         if( $a ) {
  92                                 $path = isset( $a['path'] ) ? $a['path'] : '';
  93
  94                                 global $wgScript;
  95                                 if( $path == $wgScript && $want !== 'all' ) {
  96                                         // Script inside a rewrite path?
  97                                         // Abort to keep from breaking...
  98                                         return $matches;
  99                                 }
 100
 101                                 $router = new PathRouter;
 102
 103                                 // Raw PATH_INFO style
 104                                 $router->add( "$wgScript/$1" );
 105
 106                                 if( isset( $_SERVER['SCRIPT_NAME'] )
 107                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] ) )
 108                                 {
 109                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 110                                         # But we do have some other .php files such as img_auth.php
 111                                         # Don't let root article paths clober the parsing for them
 112                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 113                                 }
 114
 115                                 global $wgArticlePath;
 116                                 if( $wgArticlePath ) {
 117                                         $router->add( $wgArticlePath );
 118                                 }
 119
 120                                 global $wgActionPaths;
 121                                 if( $wgActionPaths ) {
 122                                         $router->add( $wgActionPaths, array( 'action' => '$key' ) );
 123                                 }
 124
 125                                 global $wgVariantArticlePath, $wgContLang;
 126                                 if( $wgVariantArticlePath ) {
 127                                         $router->add( $wgVariantArticlePath,
 128                                                 array( 'variant' => '$2'),
 129                                                 array( '$2' => $wgContLang->getVariants() )
 130                                         );
 131                                 }
 132
 133                                 wfRunHooks( 'WebRequestPathInfoRouter', array( $router ) );
 134
 135                                 $matches = $router->parse( $path );
 136                         }
 137                 } elseif ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 138                         // Mangled PATH_INFO
 139                         // http://bugs.php.net/bug.php?id=31892
 140                         // Also reported when ini_get('cgi.fix_pathinfo')==false
 141                         $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 142
 143                 } elseif ( isset( $_SERVER['PATH_INFO'] ) && ($_SERVER['PATH_INFO'] != '') ) {
 144                         // Regular old PATH_INFO yay
 145                         $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 146                 }
 147
 148                 return $matches;
 149         }
 150
 151         /**
 152          * Work out an appropriate URL prefix containing scheme and host, based on
 153          * information detected from $_SERVER
 154          *
 155          * @return string
 156          */
 157         public static function detectServer() {
 158                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 159
 160                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 161                 $host = 'localhost';
 162                 $port = $stdPort;
 163                 foreach ( $varNames as $varName ) {
 164                         if ( !isset( $_SERVER[$varName] ) ) {
 165                                 continue;
 166                         }
 167                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 168                         if ( !$parts ) {
 169                                 // Invalid, do not use
 170                                 continue;
 171                         }
 172                         $host = $parts[0];
 173                         if ( $parts[1] === false ) {
 174                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 175                                         $port = $_SERVER['SERVER_PORT'];
 176                                 } // else leave it as $stdPort
 177                         } else {
 178                                 $port = $parts[1];
 179                         }
 180                         break;
 181                 }
 182
 183                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 184         }
 185
 186         /**
 187          * @return array
 188          */
 189         public static function detectProtocolAndStdPort() {
 190                 return ( isset( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] == 'on' ) ? array( 'https', 443 ) : array( 'http', 80 );
 191         }
 192
 193         /**
 194          * @return string
 195          */
 196         public static function detectProtocol() {
 197                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 198                 return $proto;
 199         }
 200
 201         /**
 202          * Check for title, action, and/or variant data in the URL
 203          * and interpolate it into the GET variables.
 204          * This should only be run after $wgContLang is available,
 205          * as we may need the list of language variants to determine
 206          * available variant URLs.
 207          */
 208         public function interpolateTitle() {
 209                 global $wgUsePathInfo;
 210
 211                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 212                 if ( defined( 'MW_API' ) ) {
 213                         return;
 214                 }
 215
 216                 if ( $wgUsePathInfo ) {
 217                         $matches = self::getPathInfo( 'title' );
 218                         foreach( $matches as $key => $val) {
 219                                 $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 220                         }
 221                 }
 222         }
 223
 224         /**
 225          * URL rewriting function; tries to extract page title and,
 226          * optionally, one other fixed parameter value from a URL path.
 227          *
 228          * @param $path string: the URL path given from the client
 229          * @param $bases array: one or more URLs, optionally with $1 at the end
 230          * @param $key string: if provided, the matching key in $bases will be
 231          *             passed on as the value of this URL parameter
 232          * @return array of URL variables to interpolate; empty if no match
 233          */
 234         static function extractTitle( $path, $bases, $key = false ) {
 235                 foreach( (array)$bases as $keyValue => $base ) {
 236                         // Find the part after $wgArticlePath
 237                         $base = str_replace( '$1', '', $base );
 238                         $baseLen = strlen( $base );
 239                         if( substr( $path, 0, $baseLen ) == $base ) {
 240                                 $raw = substr( $path, $baseLen );
 241                                 if( $raw !== '' ) {
 242                                         $matches = array( 'title' => rawurldecode( $raw ) );
 243                                         if( $key ) {
 244                                                 $matches[$key] = $keyValue;
 245                                         }
 246                                         return $matches;
 247                                 }
 248                         }
 249                 }
 250                 return array();
 251         }
 252
 253         /**
 254          * Recursively strips slashes from the given array;
 255          * used for undoing the evil that is magic_quotes_gpc.
 256          *
 257          * @param $arr array: will be modified
 258          * @param $topLevel bool Specifies if the array passed is from the top
 259          * level of the source. In PHP5 magic_quotes only escapes the first level
 260          * of keys that belong to an array.
 261          * @return array the original array
 262          * @see http://www.php.net/manual/en/function.get-magic-quotes-gpc.php#49612
 263          */
 264         private function &fix_magic_quotes( &$arr, $topLevel = true ) {
 265                 $clean = array();
 266                 foreach( $arr as $key => $val ) {
 267                         if( is_array( $val ) ) {
 268                                 $cleanKey = $topLevel ? stripslashes( $key ) : $key;
 269                                 $clean[$cleanKey] = $this->fix_magic_quotes( $arr[$key], false );
 270                         } else {
 271                                 $cleanKey = stripslashes( $key );
 272                                 $clean[$cleanKey] = stripslashes( $val );
 273                         }
 274                 }
 275                 $arr = $clean;
 276                 return $arr;
 277         }
 278
 279         /**
 280          * If magic_quotes_gpc option is on, run the global arrays
 281          * through fix_magic_quotes to strip out the stupid slashes.
 282          * WARNING: This should only be done once! Running a second
 283          * time could damage the values.
 284          */
 285         private function checkMagicQuotes() {
 286                 $mustFixQuotes = function_exists( 'get_magic_quotes_gpc' )
 287                         && get_magic_quotes_gpc();
 288                 if( $mustFixQuotes ) {
 289                         $this->fix_magic_quotes( $_COOKIE );
 290                         $this->fix_magic_quotes( $_ENV );
 291                         $this->fix_magic_quotes( $_GET );
 292                         $this->fix_magic_quotes( $_POST );
 293                         $this->fix_magic_quotes( $_REQUEST );
 294                         $this->fix_magic_quotes( $_SERVER );
 295                 }
 296         }
 297
 298         /**
 299          * Recursively normalizes UTF-8 strings in the given array.
 300          *
 301          * @param $data string or array
 302          * @return cleaned-up version of the given
 303          * @private
 304          */
 305         function normalizeUnicode( $data ) {
 306                 if( is_array( $data ) ) {
 307                         foreach( $data as $key => $val ) {
 308                                 $data[$key] = $this->normalizeUnicode( $val );
 309                         }
 310                 } else {
 311                         global $wgContLang;
 312                         $data = isset( $wgContLang ) ? $wgContLang->normalize( $data ) : UtfNormal::cleanUp( $data );
 313                 }
 314                 return $data;
 315         }
 316
 317         /**
 318          * Fetch a value from the given array or return $default if it's not set.
 319          *
 320          * @param $arr Array
 321          * @param $name String
 322          * @param $default Mixed
 323          * @return mixed
 324          */
 325         private function getGPCVal( $arr, $name, $default ) {
 326                 # PHP is so nice to not touch input data, except sometimes:
 327                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 328                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 329                 $name = strtr( $name, '.', '_' );
 330                 if( isset( $arr[$name] ) ) {
 331                         global $wgContLang;
 332                         $data = $arr[$name];
 333                         if( isset( $_GET[$name] ) && !is_array( $data ) ) {
 334                                 # Check for alternate/legacy character encoding.
 335                                 if( isset( $wgContLang ) ) {
 336                                         $data = $wgContLang->checkTitleEncoding( $data );
 337                                 }
 338                         }
 339                         $data = $this->normalizeUnicode( $data );
 340                         return $data;
 341                 } else {
 342                         taint( $default );
 343                         return $default;
 344                 }
 345         }
 346
 347         /**
 348          * Fetch a scalar from the input or return $default if it's not set.
 349          * Returns a string. Arrays are discarded. Useful for
 350          * non-freeform text inputs (e.g. predefined internal text keys
 351          * selected by a drop-down menu). For freeform input, see getText().
 352          *
 353          * @param $name String
 354          * @param $default String: optional default (or NULL)
 355          * @return String
 356          */
 357         public function getVal( $name, $default = null ) {
 358                 $val = $this->getGPCVal( $this->data, $name, $default );
 359                 if( is_array( $val ) ) {
 360                         $val = $default;
 361                 }
 362                 if( is_null( $val ) ) {
 363                         return $val;
 364                 } else {
 365                         return (string)$val;
 366                 }
 367         }
 368
 369         /**
 370          * Set an arbitrary value into our get/post data.
 371          *
 372          * @param $key String: key name to use
 373          * @param $value Mixed: value to set
 374          * @return Mixed: old value if one was present, null otherwise
 375          */
 376         public function setVal( $key, $value ) {
 377                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 378                 $this->data[$key] = $value;
 379                 return $ret;
 380         }
 381
 382         /**
 383          * Fetch an array from the input or return $default if it's not set.
 384          * If source was scalar, will return an array with a single element.
 385          * If no source and no default, returns NULL.
 386          *
 387          * @param $name String
 388          * @param $default Array: optional default (or NULL)
 389          * @return Array
 390          */
 391         public function getArray( $name, $default = null ) {
 392                 $val = $this->getGPCVal( $this->data, $name, $default );
 393                 if( is_null( $val ) ) {
 394                         return null;
 395                 } else {
 396                         return (array)$val;
 397                 }
 398         }
 399
 400         /**
 401          * Fetch an array of integers, or return $default if it's not set.
 402          * If source was scalar, will return an array with a single element.
 403          * If no source and no default, returns NULL.
 404          * If an array is returned, contents are guaranteed to be integers.
 405          *
 406          * @param $name String
 407          * @param $default Array: option default (or NULL)
 408          * @return Array of ints
 409          */
 410         public function getIntArray( $name, $default = null ) {
 411                 $val = $this->getArray( $name, $default );
 412                 if( is_array( $val ) ) {
 413                         $val = array_map( 'intval', $val );
 414                 }
 415                 return $val;
 416         }
 417
 418         /**
 419          * Fetch an integer value from the input or return $default if not set.
 420          * Guaranteed to return an integer; non-numeric input will typically
 421          * return 0.
 422          *
 423          * @param $name String
 424          * @param $default Integer
 425          * @return Integer
 426          */
 427         public function getInt( $name, $default = 0 ) {
 428                 return intval( $this->getVal( $name, $default ) );
 429         }
 430
 431         /**
 432          * Fetch an integer value from the input or return null if empty.
 433          * Guaranteed to return an integer or null; non-numeric input will
 434          * typically return null.
 435          *
 436          * @param $name String
 437          * @return Integer
 438          */
 439         public function getIntOrNull( $name ) {
 440                 $val = $this->getVal( $name );
 441                 return is_numeric( $val )
 442                         ? intval( $val )
 443                         : null;
 444         }
 445
 446         /**
 447          * Fetch a boolean value from the input or return $default if not set.
 448          * Guaranteed to return true or false, with normal PHP semantics for
 449          * boolean interpretation of strings.
 450          *
 451          * @param $name String
 452          * @param $default Boolean
 453          * @return Boolean
 454          */
 455         public function getBool( $name, $default = false ) {
 456                 return (bool)$this->getVal( $name, $default );
 457         }
 458
 459         /**
 460          * Fetch a boolean value from the input or return $default if not set.
 461          * Unlike getBool, the string "false" will result in boolean false, which is
 462          * useful when interpreting information sent from JavaScript.
 463          *
 464          * @param $name String
 465          * @param $default Boolean
 466          * @return Boolean
 467          */
 468         public function getFuzzyBool( $name, $default = false ) {
 469                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 470         }
 471
 472         /**
 473          * Return true if the named value is set in the input, whatever that
 474          * value is (even "0"). Return false if the named value is not set.
 475          * Example use is checking for the presence of check boxes in forms.
 476          *
 477          * @param $name String
 478          * @return Boolean
 479          */
 480         public function getCheck( $name ) {
 481                 # Checkboxes and buttons are only present when clicked
 482                 # Presence connotes truth, abscense false
 483                 $val = $this->getVal( $name, null );
 484                 return isset( $val );
 485         }
 486
 487         /**
 488          * Fetch a text string from the given array or return $default if it's not
 489          * set. Carriage returns are stripped from the text, and with some language
 490          * modules there is an input transliteration applied. This should generally
 491          * be used for form <textarea> and <input> fields. Used for user-supplied
 492          * freeform text input (for which input transformations may be required - e.g.
 493          * Esperanto x-coding).
 494          *
 495          * @param $name String
 496          * @param $default String: optional
 497          * @return String
 498          */
 499         public function getText( $name, $default = '' ) {
 500                 global $wgContLang;
 501                 $val = $this->getVal( $name, $default );
 502                 return str_replace( "\r\n", "\n",
 503                         $wgContLang->recodeInput( $val ) );
 504         }
 505
 506         /**
 507          * Extracts the given named values into an array.
 508          * If no arguments are given, returns all input values.
 509          * No transformation is performed on the values.
 510          *
 511          * @return array
 512          */
 513         public function getValues() {
 514                 $names = func_get_args();
 515                 if ( count( $names ) == 0 ) {
 516                         $names = array_keys( $this->data );
 517                 }
 518
 519                 $retVal = array();
 520                 foreach ( $names as $name ) {
 521                         $value = $this->getVal( $name );
 522                         if ( !is_null( $value ) ) {
 523                                 $retVal[$name] = $value;
 524                         }
 525                 }
 526                 return $retVal;
 527         }
 528
 529         /**
 530          * Returns the names of all input values excluding those in $exclude.
 531          *
 532          * @param $exclude Array
 533          * @return array
 534          */
 535         public function getValueNames( $exclude = array() ) {
 536                 return array_diff( array_keys( $this->getValues() ), $exclude );
 537         }
 538
 539         /**
 540          * Get the values passed in the query string.
 541          * No transformation is performed on the values.
 542          *
 543          * @return Array
 544          */
 545          public function getQueryValues() {
 546                 return $_GET;
 547          }
 548
 549         /**
 550          * Returns true if the present request was reached by a POST operation,
 551          * false otherwise (GET, HEAD, or command-line).
 552          *
 553          * Note that values retrieved by the object may come from the
 554          * GET URL etc even on a POST request.
 555          *
 556          * @return Boolean
 557          */
 558         public function wasPosted() {
 559                 return isset( $_SERVER['REQUEST_METHOD'] ) && $_SERVER['REQUEST_METHOD'] == 'POST';
 560         }
 561
 562         /**
 563          * Returns true if there is a session cookie set.
 564          * This does not necessarily mean that the user is logged in!
 565          *
 566          * If you want to check for an open session, use session_id()
 567          * instead; that will also tell you if the session was opened
 568          * during the current request (in which case the cookie will
 569          * be sent back to the client at the end of the script run).
 570          *
 571          * @return Boolean
 572          */
 573         public function checkSessionCookie() {
 574                 return isset( $_COOKIE[ session_name() ] );
 575         }
 576
 577         /**
 578          * Get a cookie from the $_COOKIE jar
 579          *
 580          * @param $key String: the name of the cookie
 581          * @param $prefix String: a prefix to use for the cookie name, if not $wgCookiePrefix
 582          * @param $default Mixed: what to return if the value isn't found
 583          * @return Mixed: cookie value or $default if the cookie not set
 584          */
 585         public function getCookie( $key, $prefix = null, $default = null ) {
 586                 if( $prefix === null ) {
 587                         global $wgCookiePrefix;
 588                         $prefix = $wgCookiePrefix;
 589                 }
 590                 return $this->getGPCVal( $_COOKIE, $prefix . $key , $default );
 591         }
 592
 593         /**
 594          * Return the path and query string portion of the request URI.
 595          * This will be suitable for use as a relative link in HTML output.
 596          *
 597          * @return String
 598          */
 599         public function getRequestURL() {
 600                 if( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 601                         $base = $_SERVER['REQUEST_URI'];
 602                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] ) && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] ) ) {
 603                         // Probably IIS; doesn't set REQUEST_URI
 604                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 605                 } elseif( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 606                         $base = $_SERVER['SCRIPT_NAME'];
 607                         if( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 608                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 609                         }
 610                 } else {
 611                         // This shouldn't happen!
 612                         throw new MWException( "Web server doesn't provide either " .
 613                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 614                                 "of your web server configuration to http://bugzilla.wikimedia.org/" );
 615                 }
 616                 // User-agents should not send a fragment with the URI, but
 617                 // if they do, and the web server passes it on to us, we
 618                 // need to strip it or we get false-positive redirect loops
 619                 // or weird output URLs
 620                 $hash = strpos( $base, '#' );
 621                 if( $hash !== false ) {
 622                         $base = substr( $base, 0, $hash );
 623                 }
 624                 if( $base[0] == '/' ) {
 625                         return $base;
 626                 } else {
 627                         // We may get paths with a host prepended; strip it.
 628                         return preg_replace( '!^[^:]+://[^/]+/!', '/', $base );
 629                 }
 630         }
 631
 632         /**
 633          * Return the request URI with the canonical service and hostname, path,
 634          * and query string. This will be suitable for use as an absolute link
 635          * in HTML or other output.
 636          *
 637          * If $wgServer is protocol-relative, this will return a fully
 638          * qualified URL with the protocol that was used for this request.
 639          *
 640          * @return String
 641          */
 642         public function getFullRequestURL() {
 643                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 644         }
 645
 646         /**
 647          * Take an arbitrary query and rewrite the present URL to include it
 648          * @param $query String: query string fragment; do not include initial '?'
 649          *
 650          * @return String
 651          */
 652         public function appendQuery( $query ) {
 653                 return $this->appendQueryArray( wfCgiToArray( $query ) );
 654         }
 655
 656         /**
 657          * HTML-safe version of appendQuery().
 658          *
 659          * @param $query String: query string fragment; do not include initial '?'
 660          * @return String
 661          */
 662         public function escapeAppendQuery( $query ) {
 663                 return htmlspecialchars( $this->appendQuery( $query ) );
 664         }
 665
 666         /**
 667          * @param $key
 668          * @param $value
 669          * @param $onlyquery bool
 670          * @return String
 671          */
 672         public function appendQueryValue( $key, $value, $onlyquery = false ) {
 673                 return $this->appendQueryArray( array( $key => $value ), $onlyquery );
 674         }
 675
 676         /**
 677          * Appends or replaces value of query variables.
 678          *
 679          * @param $array Array of values to replace/add to query
 680          * @param $onlyquery Bool: whether to only return the query string and not
 681          *                   the complete URL
 682          * @return String
 683          */
 684         public function appendQueryArray( $array, $onlyquery = false ) {
 685                 global $wgTitle;
 686                 $newquery = $this->getQueryValues();
 687                 unset( $newquery['title'] );
 688                 $newquery = array_merge( $newquery, $array );
 689                 $query = wfArrayToCGI( $newquery );
 690                 return $onlyquery ? $query : $wgTitle->getLocalURL( $query );
 691         }
 692
 693         /**
 694          * Check for limit and offset parameters on the input, and return sensible
 695          * defaults if not given. The limit must be positive and is capped at 5000.
 696          * Offset must be positive but is not capped.
 697          *
 698          * @param $deflimit Integer: limit to use if no input and the user hasn't set the option.
 699          * @param $optionname String: to specify an option other than rclimit to pull from.
 700          * @return array first element is limit, second is offset
 701          */
 702         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 703                 global $wgUser;
 704
 705                 $limit = $this->getInt( 'limit', 0 );
 706                 if( $limit < 0 ) {
 707                         $limit = 0;
 708                 }
 709                 if( ( $limit == 0 ) && ( $optionname != '' ) ) {
 710                         $limit = (int)$wgUser->getOption( $optionname );
 711                 }
 712                 if( $limit <= 0 ) {
 713                         $limit = $deflimit;
 714                 }
 715                 if( $limit > 5000 ) {
 716                         $limit = 5000; # We have *some* limits...
 717                 }
 718
 719                 $offset = $this->getInt( 'offset', 0 );
 720                 if( $offset < 0 ) {
 721                         $offset = 0;
 722                 }
 723
 724                 return array( $limit, $offset );
 725         }
 726
 727         /**
 728          * Return the path to the temporary file where PHP has stored the upload.
 729          *
 730          * @param $key String:
 731          * @return string or NULL if no such file.
 732          */
 733         public function getFileTempname( $key ) {
 734                 $file = new WebRequestUpload( $this, $key );
 735                 return $file->getTempName();
 736         }
 737
 738         /**
 739          * Return the size of the upload, or 0.
 740          *
 741          * @deprecated since 1.17
 742          * @param $key String:
 743          * @return integer
 744          */
 745         public function getFileSize( $key ) {
 746                 wfDeprecated( __METHOD__, '1.17' );
 747                 $file = new WebRequestUpload( $this, $key );
 748                 return $file->getSize();
 749         }
 750
 751         /**
 752          * Return the upload error or 0
 753          *
 754          * @param $key String:
 755          * @return integer
 756          */
 757         public function getUploadError( $key ) {
 758                 $file = new WebRequestUpload( $this, $key );
 759                 return $file->getError();
 760         }
 761
 762         /**
 763          * Return the original filename of the uploaded file, as reported by
 764          * the submitting user agent. HTML-style character entities are
 765          * interpreted and normalized to Unicode normalization form C, in part
 766          * to deal with weird input from Safari with non-ASCII filenames.
 767          *
 768          * Other than this the name is not verified for being a safe filename.
 769          *
 770          * @param $key String:
 771          * @return string or NULL if no such file.
 772          */
 773         public function getFileName( $key ) {
 774                 $file = new WebRequestUpload( $this, $key );
 775                 return $file->getName();
 776         }
 777
 778         /**
 779          * Return a WebRequestUpload object corresponding to the key
 780          *
 781          * @param $key string
 782          * @return WebRequestUpload
 783          */
 784         public function getUpload( $key ) {
 785                 return new WebRequestUpload( $this, $key );
 786         }
 787
 788         /**
 789          * Return a handle to WebResponse style object, for setting cookies,
 790          * headers and other stuff, for Request being worked on.
 791          *
 792          * @return WebResponse
 793          */
 794         public function response() {
 795                 /* Lazy initialization of response object for this request */
 796                 if ( !is_object( $this->response ) ) {
 797                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 798                         $this->response = new $class();
 799                 }
 800                 return $this->response;
 801         }
 802
 803         /**
 804          * Initialise the header list
 805          */
 806         private function initHeaders() {
 807                 if ( count( $this->headers ) ) {
 808                         return;
 809                 }
 810
 811                 if ( function_exists( 'apache_request_headers' ) ) {
 812                         foreach ( apache_request_headers() as $tempName => $tempValue ) {
 813                                 $this->headers[ strtoupper( $tempName ) ] = $tempValue;
 814                         }
 815                 } else {
 816                         foreach ( $_SERVER as $name => $value ) {
 817                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 818                                         $name = str_replace( '_', '-',  substr( $name, 5 ) );
 819                                         $this->headers[$name] = $value;
 820                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 821                                         $this->headers['CONTENT-LENGTH'] = $value;
 822                                 }
 823                         }
 824                 }
 825         }
 826
 827         /**
 828          * Get an array containing all request headers
 829          *
 830          * @return Array mapping header name to its value
 831          */
 832         public function getAllHeaders() {
 833                 $this->initHeaders();
 834                 return $this->headers;
 835         }
 836
 837         /**
 838          * Get a request header, or false if it isn't set
 839          * @param $name String: case-insensitive header name
 840          *
 841          * @return string|false
 842          */
 843         public function getHeader( $name ) {
 844                 $this->initHeaders();
 845                 $name = strtoupper( $name );
 846                 if ( isset( $this->headers[$name] ) ) {
 847                         return $this->headers[$name];
 848                 } else {
 849                         return false;
 850                 }
 851         }
 852
 853         /**
 854          * Get data from $_SESSION
 855          *
 856          * @param $key String: name of key in $_SESSION
 857          * @return Mixed
 858          */
 859         public function getSessionData( $key ) {
 860                 if( !isset( $_SESSION[$key] ) ) {
 861                         return null;
 862                 }
 863                 return $_SESSION[$key];
 864         }
 865
 866         /**
 867          * Set session data
 868          *
 869          * @param $key String: name of key in $_SESSION
 870          * @param $data Mixed
 871          */
 872         public function setSessionData( $key, $data ) {
 873                 $_SESSION[$key] = $data;
 874         }
 875
 876         /**
 877          * Check if Internet Explorer will detect an incorrect cache extension in
 878          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 879          * message or redirect to a safer URL. Returns true if the URL is OK, and
 880          * false if an error message has been shown and the request should be aborted.
 881          *
 882          * @param $extWhitelist array
 883          * @return bool
 884          */
 885         public function checkUrlExtension( $extWhitelist = array() ) {
 886                 global $wgScriptExtension;
 887                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 888                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 889                         if ( !$this->wasPosted() ) {
 890                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 891                                         $this->getFullRequestURL(), $extWhitelist );
 892                                 if ( $newUrl !== false ) {
 893                                         $this->doSecurityRedirect( $newUrl );
 894                                         return false;
 895                                 }
 896                         }
 897                         throw new HttpError( 403,
 898                                 'Invalid file extension found in the path info or query string.' );
 899                 }
 900                 return true;
 901         }
 902
 903         /**
 904          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 905          * IE 6. Returns true if it was successful, false otherwise.
 906          *
 907          * @param $url string
 908          * @return bool
 909          */
 910         protected function doSecurityRedirect( $url ) {
 911                 header( 'Location: ' . $url );
 912                 header( 'Content-Type: text/html' );
 913                 $encUrl = htmlspecialchars( $url );
 914                 echo <<<HTML
 915 <html>
 916 <head>
 917 <title>Security redirect</title>
 918 </head>
 919 <body>
 920 <h1>Security redirect</h1>
 921 <p>
 922 We can't serve non-HTML content from the URL you have requested, because
 923 Internet Explorer would interpret it as an incorrect and potentially dangerous
 924 content type.</p>
 925 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the URL you have requested, except that
 926 "&amp;*" is appended. This prevents Internet Explorer from seeing a bogus file
 927 extension.
 928 </p>
 929 </body>
 930 </html>
 931 HTML;
 932                 echo "\n";
 933                 return true;
 934         }
 935
 936         /**
 937          * Returns true if the PATH_INFO ends with an extension other than a script
 938          * extension. This could confuse IE for scripts that send arbitrary data which
 939          * is not HTML but may be detected as such.
 940          *
 941          * Various past attempts to use the URL to make this check have generally
 942          * run up against the fact that CGI does not provide a standard method to
 943          * determine the URL. PATH_INFO may be mangled (e.g. if cgi.fix_pathinfo=0),
 944          * but only by prefixing it with the script name and maybe some other stuff,
 945          * the extension is not mangled. So this should be a reasonably portable
 946          * way to perform this security check.
 947          *
 948          * Also checks for anything that looks like a file extension at the end of
 949          * QUERY_STRING, since IE 6 and earlier will use this to get the file type
 950          * if there was no dot before the question mark (bug 28235).
 951          *
 952          * @deprecated Use checkUrlExtension().
 953          *
 954          * @param $extWhitelist array
 955          *
 956          * @return bool
 957          */
 958         public function isPathInfoBad( $extWhitelist = array() ) {
 959                 wfDeprecated( __METHOD__, '1.17' );
 960                 global $wgScriptExtension;
 961                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 962                 return IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist );
 963         }
 964
 965         /**
 966          * Parse the Accept-Language header sent by the client into an array
 967          * @return array array( languageCode => q-value ) sorted by q-value in descending order
 968          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 969          * This is aligned with rfc2616 section 14.4
 970          */
 971         public function getAcceptLang() {
 972                 // Modified version of code found at http://www.thefutureoftheweb.com/blog/use-accept-language-header
 973                 $acceptLang = $this->getHeader( 'Accept-Language' );
 974                 if ( !$acceptLang ) {
 975                         return array();
 976                 }
 977
 978                 // Return the language codes in lower case
 979                 $acceptLang = strtolower( $acceptLang );
 980
 981                 // Break up string into pieces (languages and q factors)
 982                 $lang_parse = null;
 983                 preg_match_all( '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
 984                         $acceptLang, $lang_parse );
 985
 986                 if ( !count( $lang_parse[1] ) ) {
 987                         return array();
 988                 }
 989
 990                 // Create a list like "en" => 0.8
 991                 $langs = array_combine( $lang_parse[1], $lang_parse[4] );
 992                 // Set default q factor to 1
 993                 foreach ( $langs as $lang => $val ) {
 994                         if ( $val === '' ) {
 995                                 $langs[$lang] = 1;
 996                         } elseif ( $val == 0 ) {
 997                                 unset($langs[$lang]);
 998                         }
 999                 }
1000
1001                 // Sort list
1002                 arsort( $langs, SORT_NUMERIC );
1003                 return $langs;
1004         }
1005
1006         /**
1007          * Fetch the raw IP from the request
1008          *
1009          * @since 1.19
1010          *
1011          * @return String
1012          */
1013         protected function getRawIP() {
1014                 if ( isset( $_SERVER['REMOTE_ADDR'] ) ) {
1015                         return IP::canonicalize( $_SERVER['REMOTE_ADDR'] );
1016                 } else {
1017                         return null;
1018                 }
1019         }
1020
1021         /**
1022          * Work out the IP address based on various globals
1023          * For trusted proxies, use the XFF client IP (first of the chain)
1024          *
1025          * @since 1.19
1026          *
1027          * @return string
1028          */
1029         public function getIP() {
1030                 global $wgUsePrivateIPs;
1031
1032                 # Return cached result
1033                 if ( $this->ip !== null ) {
1034                         return $this->ip;
1035                 }
1036
1037                 # collect the originating ips
1038                 $ip = $this->getRawIP();
1039
1040                 # Append XFF
1041                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1042                 if ( $forwardedFor !== false ) {
1043                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1044                         $ipchain = array_reverse( $ipchain );
1045                         if ( $ip ) {
1046                                 array_unshift( $ipchain, $ip );
1047                         }
1048
1049                         # Step through XFF list and find the last address in the list which is a trusted server
1050                         # Set $ip to the IP address given by that trusted server, unless the address is not sensible (e.g. private)
1051                         foreach ( $ipchain as $i => $curIP ) {
1052                                 $curIP = IP::canonicalize( $curIP );
1053                                 if ( wfIsTrustedProxy( $curIP ) ) {
1054                                         if ( isset( $ipchain[$i + 1] ) ) {
1055                                                 if ( $wgUsePrivateIPs || IP::isPublic( $ipchain[$i + 1 ] ) ) {
1056                                                         $ip = $ipchain[$i + 1];
1057                                                 }
1058                                         }
1059                                 } else {
1060                                         break;
1061                                 }
1062                         }
1063                 }
1064
1065                 # Allow extensions to improve our guess
1066                 wfRunHooks( 'GetIP', array( &$ip ) );
1067
1068                 if ( !$ip ) {
1069                         throw new MWException( "Unable to determine IP" );
1070                 }
1071
1072                 wfDebug( "IP: $ip\n" );
1073                 $this->ip = $ip;
1074                 return $ip;
1075         }
1076 }
1077
1078 /**
1079  * Object to access the $_FILES array
1080  */
1081 class WebRequestUpload {
1082         protected $request;
1083         protected $doesExist;
1084         protected $fileInfo;
1085
1086         /**
1087          * Constructor. Should only be called by WebRequest
1088          *
1089          * @param $request WebRequest The associated request
1090          * @param $key string Key in $_FILES array (name of form field)
1091          */
1092         public function __construct( $request, $key ) {
1093                 $this->request = $request;
1094                 $this->doesExist = isset( $_FILES[$key] );
1095                 if ( $this->doesExist ) {
1096                         $this->fileInfo = $_FILES[$key];
1097                 }
1098         }
1099
1100         /**
1101          * Return whether a file with this name was uploaded.
1102          *
1103          * @return bool
1104          */
1105         public function exists() {
1106                 return $this->doesExist;
1107         }
1108
1109         /**
1110          * Return the original filename of the uploaded file
1111          *
1112          * @return mixed Filename or null if non-existent
1113          */
1114         public function getName() {
1115                 if ( !$this->exists() ) {
1116                         return null;
1117                 }
1118
1119                 global $wgContLang;
1120                 $name = $this->fileInfo['name'];
1121
1122                 # Safari sends filenames in HTML-encoded Unicode form D...
1123                 # Horrid and evil! Let's try to make some kind of sense of it.
1124                 $name = Sanitizer::decodeCharReferences( $name );
1125                 $name = $wgContLang->normalize( $name );
1126                 wfDebug( __METHOD__ . ": {$this->fileInfo['name']} normalized to '$name'\n" );
1127                 return $name;
1128         }
1129
1130         /**
1131          * Return the file size of the uploaded file
1132          *
1133          * @return int File size or zero if non-existent
1134          */
1135         public function getSize() {
1136                 if ( !$this->exists() ) {
1137                         return 0;
1138                 }
1139
1140                 return $this->fileInfo['size'];
1141         }
1142
1143         /**
1144          * Return the path to the temporary file
1145          *
1146          * @return mixed Path or null if non-existent
1147          */
1148         public function getTempName() {
1149                 if ( !$this->exists() ) {
1150                         return null;
1151                 }
1152
1153                 return $this->fileInfo['tmp_name'];
1154         }
1155
1156         /**
1157          * Return the upload error. See link for explanation
1158          * http://www.php.net/manual/en/features.file-upload.errors.php
1159          *
1160          * @return int One of the UPLOAD_ constants, 0 if non-existent
1161          */
1162         public function getError() {
1163                 if ( !$this->exists() ) {
1164                         return 0; # UPLOAD_ERR_OK
1165                 }
1166
1167                 return $this->fileInfo['error'];
1168         }
1169
1170         /**
1171          * Returns whether this upload failed because of overflow of a maximum set
1172          * in php.ini
1173          *
1174          * @return bool
1175          */
1176         public function isIniSizeOverflow() {
1177                 if ( $this->getError() == UPLOAD_ERR_INI_SIZE ) {
1178                         # PHP indicated that upload_max_filesize is exceeded
1179                         return true;
1180                 }
1181
1182                 $contentLength = $this->request->getHeader( 'CONTENT_LENGTH' );
1183                 if ( $contentLength > wfShorthandToInteger( ini_get( 'post_max_size' ) ) ) {
1184                         # post_max_size is exceeded
1185                         return true;
1186                 }
1187
1188                 return false;
1189         }
1190 }
1191
1192 /**
1193  * WebRequest clone which takes values from a provided array.
1194  *
1195  * @ingroup HTTP
1196  */
1197 class FauxRequest extends WebRequest {
1198         private $wasPosted = false;
1199         private $session = array();
1200
1201         /**
1202          * @param $data Array of *non*-urlencoded key => value pairs, the
1203          *   fake GET/POST values
1204          * @param $wasPosted Bool: whether to treat the data as POST
1205          * @param $session Mixed: session array or null
1206          */
1207         public function __construct( $data = array(), $wasPosted = false, $session = null ) {
1208                 if( is_array( $data ) ) {
1209                         $this->data = $data;
1210                 } else {
1211                         throw new MWException( "FauxRequest() got bogus data" );
1212                 }
1213                 $this->wasPosted = $wasPosted;
1214                 if( $session )
1215                         $this->session = $session;
1216         }
1217
1218         /**
1219          * @param $method string
1220          * @throws MWException
1221          */
1222         private function notImplemented( $method ) {
1223                 throw new MWException( "{$method}() not implemented" );
1224         }
1225
1226         /**
1227          * @param $name string
1228          * @param $default string
1229          * @return string
1230          */
1231         public function getText( $name, $default = '' ) {
1232                 # Override; don't recode since we're using internal data
1233                 return (string)$this->getVal( $name, $default );
1234         }
1235
1236         /**
1237          * @return Array
1238          */
1239         public function getValues() {
1240                 return $this->data;
1241         }
1242
1243         /**
1244          * @return array
1245          */
1246         public function getQueryValues() {
1247                 if ( $this->wasPosted ) {
1248                         return array();
1249                 } else {
1250                         return $this->data;
1251                 }
1252         }
1253
1254         /**
1255          * @return bool
1256          */
1257         public function wasPosted() {
1258                 return $this->wasPosted;
1259         }
1260
1261         public function checkSessionCookie() {
1262                 return false;
1263         }
1264
1265         public function getRequestURL() {
1266                 $this->notImplemented( __METHOD__ );
1267         }
1268
1269         /**
1270          * @param $name
1271          * @return bool|string
1272          */
1273         public function getHeader( $name ) {
1274                 return isset( $this->headers[$name] ) ? $this->headers[$name] : false;
1275         }
1276
1277         /**
1278          * @param $name string
1279          * @param $val string
1280          */
1281         public function setHeader( $name, $val ) {
1282                 $this->headers[$name] = $val;
1283         }
1284
1285         /**
1286          * @param $key
1287          * @return mixed
1288          */
1289         public function getSessionData( $key ) {
1290                 if( isset( $this->session[$key] ) )
1291                         return $this->session[$key];
1292         }
1293
1294         /**
1295          * @param $key
1296          * @param $data
1297          */
1298         public function setSessionData( $key, $data ) {
1299                 $this->session[$key] = $data;
1300         }
1301
1302         /**
1303          * @return array|Mixed|null
1304          */
1305         public function getSessionArray() {
1306                 return $this->session;
1307         }
1308
1309         /**
1310          * @param array $extWhitelist
1311          * @return bool
1312          */
1313         public function isPathInfoBad( $extWhitelist = array() ) {
1314                 return false;
1315         }
1316
1317         /**
1318          * @param array $extWhitelist
1319          * @return bool
1320          */
1321         public function checkUrlExtension( $extWhitelist = array() ) {
1322                 return true;
1323         }
1324
1325         /**
1326          * @return string
1327          */
1328         protected function getRawIP() {
1329                 return '127.0.0.1';
1330         }
1331 }
1332
1333 /**
1334  * Similar to FauxRequest, but only fakes URL parameters and method
1335  * (POST or GET) and use the base request for the remaining stuff
1336  * (cookies, session and headers).
1337  *
1338  * @ingroup HTTP
1339  */
1340 class DerivativeRequest extends FauxRequest {
1341         private $base;
1342
1343         public function __construct( WebRequest $base, $data, $wasPosted = false ) {
1344                 $this->base = $base;
1345                 parent::__construct( $data, $wasPosted );
1346         }
1347
1348         public function getCookie( $key, $prefix = null, $default = null ) {
1349                 return $this->base->getCookie( $key, $prefix, $default );
1350         }
1351
1352         public function checkSessionCookie() {
1353                 return $this->base->checkSessionCookie();
1354         }
1355
1356         public function getHeader( $name ) {
1357                 return $this->base->getHeader( $name );
1358         }
1359
1360         public function getAllHeaders() {
1361                 return $this->base->getAllHeaders();
1362         }
1363
1364         public function getSessionData( $key ) {
1365                 return $this->base->getSessionData( $key );
1366         }
1367
1368         public function setSessionData( $key, $data ) {
1369                 return $this->base->setSessionData( $key, $data );
1370         }
1371
1372         public function getAcceptLang() {
1373                 return $this->base->getAcceptLang();
1374         }
1375
1376         public function getIP() {
1377                 return $this->base->getIP();
1378         }
1379 }