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                 global $wgUsePathInfo;
  81                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
  82                 // And also by Apache 2.x, double slashes are converted to single slashes.
  83                 // So we will use REQUEST_URI if possible.
  84                 $matches = array();
  85                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
  86                         // Slurp out the path portion to examine...
  87                         $url = $_SERVER['REQUEST_URI'];
  88                         if ( !preg_match( '!^https?://!', $url ) ) {
  89                                 $url = 'http://unused' . $url;
  90                         }
  91                         wfSuppressWarnings();
  92                         $a = parse_url( $url );
  93                         wfRestoreWarnings();
  94                         if( $a ) {
  95                                 $path = isset( $a['path'] ) ? $a['path'] : '';
  96
  97                                 global $wgScript;
  98                                 if( $path == $wgScript && $want !== 'all' ) {
  99                                         // Script inside a rewrite path?
 100                                         // Abort to keep from breaking...
 101                                         return $matches;
 102                                 }
 103
 104                                 $router = new PathRouter;
 105
 106                                 // Raw PATH_INFO style
 107                                 $router->add( "$wgScript/$1" );
 108
 109                                 if( isset( $_SERVER['SCRIPT_NAME'] )
 110                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] ) )
 111                                 {
 112                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 113                                         # But we do have some other .php files such as img_auth.php
 114                                         # Don't let root article paths clober the parsing for them
 115                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 116                                 }
 117
 118                                 global $wgArticlePath;
 119                                 if( $wgArticlePath ) {
 120                                         $router->add( $wgArticlePath );
 121                                 }
 122
 123                                 global $wgActionPaths;
 124                                 if( $wgActionPaths ) {
 125                                         $router->add( $wgActionPaths, array( 'action' => '$key' ) );
 126                                 }
 127
 128                                 global $wgVariantArticlePath, $wgContLang;
 129                                 if( $wgVariantArticlePath ) {
 130                                         $router->add( $wgVariantArticlePath,
 131                                                 array( 'variant' => '$2'),
 132                                                 array( '$2' => $wgContLang->getVariants() )
 133                                         );
 134                                 }
 135
 136                                 wfRunHooks( 'WebRequestPathInfoRouter', array( $router ) );
 137
 138                                 $matches = $router->parse( $path );
 139                         }
 140                 } elseif ( $wgUsePathInfo ) {
 141                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 142                                 // Mangled PATH_INFO
 143                                 // http://bugs.php.net/bug.php?id=31892
 144                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 145                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 146
 147                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && ($_SERVER['PATH_INFO'] != '') ) {
 148                                 // Regular old PATH_INFO yay
 149                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 150                         }
 151                 }
 152
 153                 return $matches;
 154         }
 155
 156         /**
 157          * Work out an appropriate URL prefix containing scheme and host, based on
 158          * information detected from $_SERVER
 159          *
 160          * @return string
 161          */
 162         public static function detectServer() {
 163                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 164
 165                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 166                 $host = 'localhost';
 167                 $port = $stdPort;
 168                 foreach ( $varNames as $varName ) {
 169                         if ( !isset( $_SERVER[$varName] ) ) {
 170                                 continue;
 171                         }
 172                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 173                         if ( !$parts ) {
 174                                 // Invalid, do not use
 175                                 continue;
 176                         }
 177                         $host = $parts[0];
 178                         if ( $parts[1] === false ) {
 179                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 180                                         $port = $_SERVER['SERVER_PORT'];
 181                                 } // else leave it as $stdPort
 182                         } else {
 183                                 $port = $parts[1];
 184                         }
 185                         break;
 186                 }
 187
 188                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 189         }
 190
 191         /**
 192          * @return array
 193          */
 194         public static function detectProtocolAndStdPort() {
 195                 return ( isset( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] == 'on' ) ? array( 'https', 443 ) : array( 'http', 80 );
 196         }
 197
 198         /**
 199          * @return string
 200          */
 201         public static function detectProtocol() {
 202                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 203                 return $proto;
 204         }
 205
 206         /**
 207          * Check for title, action, and/or variant data in the URL
 208          * and interpolate it into the GET variables.
 209          * This should only be run after $wgContLang is available,
 210          * as we may need the list of language variants to determine
 211          * available variant URLs.
 212          */
 213         public function interpolateTitle() {
 214                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 215                 if ( defined( 'MW_API' ) ) {
 216                         return;
 217                 }
 218
 219                 $matches = self::getPathInfo( 'title' );
 220                 foreach( $matches as $key => $val) {
 221                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 222                 }
 223         }
 224
 225         /**
 226          * URL rewriting function; tries to extract page title and,
 227          * optionally, one other fixed parameter value from a URL path.
 228          *
 229          * @param $path string: the URL path given from the client
 230          * @param $bases array: one or more URLs, optionally with $1 at the end
 231          * @param $key string: if provided, the matching key in $bases will be
 232          *             passed on as the value of this URL parameter
 233          * @return array of URL variables to interpolate; empty if no match
 234          */
 235         static function extractTitle( $path, $bases, $key = false ) {
 236                 foreach( (array)$bases as $keyValue => $base ) {
 237                         // Find the part after $wgArticlePath
 238                         $base = str_replace( '$1', '', $base );
 239                         $baseLen = strlen( $base );
 240                         if( substr( $path, 0, $baseLen ) == $base ) {
 241                                 $raw = substr( $path, $baseLen );
 242                                 if( $raw !== '' ) {
 243                                         $matches = array( 'title' => rawurldecode( $raw ) );
 244                                         if( $key ) {
 245                                                 $matches[$key] = $keyValue;
 246                                         }
 247                                         return $matches;
 248                                 }
 249                         }
 250                 }
 251                 return array();
 252         }
 253
 254         /**
 255          * Recursively strips slashes from the given array;
 256          * used for undoing the evil that is magic_quotes_gpc.
 257          *
 258          * @param $arr array: will be modified
 259          * @param $topLevel bool Specifies if the array passed is from the top
 260          * level of the source. In PHP5 magic_quotes only escapes the first level
 261          * of keys that belong to an array.
 262          * @return array the original array
 263          * @see http://www.php.net/manual/en/function.get-magic-quotes-gpc.php#49612
 264          */
 265         private function &fix_magic_quotes( &$arr, $topLevel = true ) {
 266                 $clean = array();
 267                 foreach( $arr as $key => $val ) {
 268                         if( is_array( $val ) ) {
 269                                 $cleanKey = $topLevel ? stripslashes( $key ) : $key;
 270                                 $clean[$cleanKey] = $this->fix_magic_quotes( $arr[$key], false );
 271                         } else {
 272                                 $cleanKey = stripslashes( $key );
 273                                 $clean[$cleanKey] = stripslashes( $val );
 274                         }
 275                 }
 276                 $arr = $clean;
 277                 return $arr;
 278         }
 279
 280         /**
 281          * If magic_quotes_gpc option is on, run the global arrays
 282          * through fix_magic_quotes to strip out the stupid slashes.
 283          * WARNING: This should only be done once! Running a second
 284          * time could damage the values.
 285          */
 286         private function checkMagicQuotes() {
 287                 $mustFixQuotes = function_exists( 'get_magic_quotes_gpc' )
 288                         && get_magic_quotes_gpc();
 289                 if( $mustFixQuotes ) {
 290                         $this->fix_magic_quotes( $_COOKIE );
 291                         $this->fix_magic_quotes( $_ENV );
 292                         $this->fix_magic_quotes( $_GET );
 293                         $this->fix_magic_quotes( $_POST );
 294                         $this->fix_magic_quotes( $_REQUEST );
 295                         $this->fix_magic_quotes( $_SERVER );
 296                 }
 297         }
 298
 299         /**
 300          * Recursively normalizes UTF-8 strings in the given array.
 301          *
 302          * @param $data string|array
 303          * @return array|string cleaned-up version of the given
 304          * @private
 305          */
 306         function normalizeUnicode( $data ) {
 307                 if( is_array( $data ) ) {
 308                         foreach( $data as $key => $val ) {
 309                                 $data[$key] = $this->normalizeUnicode( $val );
 310                         }
 311                 } else {
 312                         global $wgContLang;
 313                         $data = isset( $wgContLang ) ? $wgContLang->normalize( $data ) : UtfNormal::cleanUp( $data );
 314                 }
 315                 return $data;
 316         }
 317
 318         /**
 319          * Fetch a value from the given array or return $default if it's not set.
 320          *
 321          * @param $arr Array
 322          * @param $name String
 323          * @param $default Mixed
 324          * @return mixed
 325          */
 326         private function getGPCVal( $arr, $name, $default ) {
 327                 # PHP is so nice to not touch input data, except sometimes:
 328                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 329                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 330                 $name = strtr( $name, '.', '_' );
 331                 if( isset( $arr[$name] ) ) {
 332                         global $wgContLang;
 333                         $data = $arr[$name];
 334                         if( isset( $_GET[$name] ) && !is_array( $data ) ) {
 335                                 # Check for alternate/legacy character encoding.
 336                                 if( isset( $wgContLang ) ) {
 337                                         $data = $wgContLang->checkTitleEncoding( $data );
 338                                 }
 339                         }
 340                         $data = $this->normalizeUnicode( $data );
 341                         return $data;
 342                 } else {
 343                         taint( $default );
 344                         return $default;
 345                 }
 346         }
 347
 348         /**
 349          * Fetch a scalar from the input or return $default if it's not set.
 350          * Returns a string. Arrays are discarded. Useful for
 351          * non-freeform text inputs (e.g. predefined internal text keys
 352          * selected by a drop-down menu). For freeform input, see getText().
 353          *
 354          * @param $name String
 355          * @param $default String: optional default (or NULL)
 356          * @return String
 357          */
 358         public function getVal( $name, $default = null ) {
 359                 $val = $this->getGPCVal( $this->data, $name, $default );
 360                 if( is_array( $val ) ) {
 361                         $val = $default;
 362                 }
 363                 if( is_null( $val ) ) {
 364                         return $val;
 365                 } else {
 366                         return (string)$val;
 367                 }
 368         }
 369
 370         /**
 371          * Set an arbitrary value into our get/post data.
 372          *
 373          * @param $key String: key name to use
 374          * @param $value Mixed: value to set
 375          * @return Mixed: old value if one was present, null otherwise
 376          */
 377         public function setVal( $key, $value ) {
 378                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 379                 $this->data[$key] = $value;
 380                 return $ret;
 381         }
 382
 383         /**
 384          * Unset an arbitrary value from our get/post data.
 385          *
 386          * @param $key String: key name to use
 387          * @return Mixed: old value if one was present, null otherwise
 388          */
 389         public function unsetVal( $key ) {
 390                 if ( !isset( $this->data[$key] ) ) {
 391                         $ret = null;
 392                 } else {
 393                         $ret = $this->data[$key];
 394                         unset( $this->data[$key] );
 395                 }
 396                 return $ret;
 397         }
 398
 399         /**
 400          * Fetch an array from the input or return $default if it's not set.
 401          * If source was scalar, will return an array with a single element.
 402          * If no source and no default, returns NULL.
 403          *
 404          * @param $name String
 405          * @param $default Array: optional default (or NULL)
 406          * @return Array
 407          */
 408         public function getArray( $name, $default = null ) {
 409                 $val = $this->getGPCVal( $this->data, $name, $default );
 410                 if( is_null( $val ) ) {
 411                         return null;
 412                 } else {
 413                         return (array)$val;
 414                 }
 415         }
 416
 417         /**
 418          * Fetch an array of integers, or return $default if it's not set.
 419          * If source was scalar, will return an array with a single element.
 420          * If no source and no default, returns NULL.
 421          * If an array is returned, contents are guaranteed to be integers.
 422          *
 423          * @param $name String
 424          * @param $default Array: option default (or NULL)
 425          * @return Array of ints
 426          */
 427         public function getIntArray( $name, $default = null ) {
 428                 $val = $this->getArray( $name, $default );
 429                 if( is_array( $val ) ) {
 430                         $val = array_map( 'intval', $val );
 431                 }
 432                 return $val;
 433         }
 434
 435         /**
 436          * Fetch an integer value from the input or return $default if not set.
 437          * Guaranteed to return an integer; non-numeric input will typically
 438          * return 0.
 439          *
 440          * @param $name String
 441          * @param $default Integer
 442          * @return Integer
 443          */
 444         public function getInt( $name, $default = 0 ) {
 445                 return intval( $this->getVal( $name, $default ) );
 446         }
 447
 448         /**
 449          * Fetch an integer value from the input or return null if empty.
 450          * Guaranteed to return an integer or null; non-numeric input will
 451          * typically return null.
 452          *
 453          * @param $name String
 454          * @return Integer
 455          */
 456         public function getIntOrNull( $name ) {
 457                 $val = $this->getVal( $name );
 458                 return is_numeric( $val )
 459                         ? intval( $val )
 460                         : null;
 461         }
 462
 463         /**
 464          * Fetch a boolean value from the input or return $default if not set.
 465          * Guaranteed to return true or false, with normal PHP semantics for
 466          * boolean interpretation of strings.
 467          *
 468          * @param $name String
 469          * @param $default Boolean
 470          * @return Boolean
 471          */
 472         public function getBool( $name, $default = false ) {
 473                 return (bool)$this->getVal( $name, $default );
 474         }
 475
 476         /**
 477          * Fetch a boolean value from the input or return $default if not set.
 478          * Unlike getBool, the string "false" will result in boolean false, which is
 479          * useful when interpreting information sent from JavaScript.
 480          *
 481          * @param $name String
 482          * @param $default Boolean
 483          * @return Boolean
 484          */
 485         public function getFuzzyBool( $name, $default = false ) {
 486                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 487         }
 488
 489         /**
 490          * Return true if the named value is set in the input, whatever that
 491          * value is (even "0"). Return false if the named value is not set.
 492          * Example use is checking for the presence of check boxes in forms.
 493          *
 494          * @param $name String
 495          * @return Boolean
 496          */
 497         public function getCheck( $name ) {
 498                 # Checkboxes and buttons are only present when clicked
 499                 # Presence connotes truth, abscense false
 500                 return $this->getVal( $name, null ) !== null;
 501         }
 502
 503         /**
 504          * Fetch a text string from the given array or return $default if it's not
 505          * set. Carriage returns are stripped from the text, and with some language
 506          * modules there is an input transliteration applied. This should generally
 507          * be used for form "<textarea>" and "<input>" fields. Used for
 508          * user-supplied freeform text input (for which input transformations may
 509          * be required - e.g.  Esperanto x-coding).
 510          *
 511          * @param $name String
 512          * @param $default String: optional
 513          * @return String
 514          */
 515         public function getText( $name, $default = '' ) {
 516                 global $wgContLang;
 517                 $val = $this->getVal( $name, $default );
 518                 return str_replace( "\r\n", "\n",
 519                         $wgContLang->recodeInput( $val ) );
 520         }
 521
 522         /**
 523          * Extracts the given named values into an array.
 524          * If no arguments are given, returns all input values.
 525          * No transformation is performed on the values.
 526          *
 527          * @return array
 528          */
 529         public function getValues() {
 530                 $names = func_get_args();
 531                 if ( count( $names ) == 0 ) {
 532                         $names = array_keys( $this->data );
 533                 }
 534
 535                 $retVal = array();
 536                 foreach ( $names as $name ) {
 537                         $value = $this->getGPCVal( $this->data, $name, null );
 538                         if ( !is_null( $value ) ) {
 539                                 $retVal[$name] = $value;
 540                         }
 541                 }
 542                 return $retVal;
 543         }
 544
 545         /**
 546          * Returns the names of all input values excluding those in $exclude.
 547          *
 548          * @param $exclude Array
 549          * @return array
 550          */
 551         public function getValueNames( $exclude = array() ) {
 552                 return array_diff( array_keys( $this->getValues() ), $exclude );
 553         }
 554
 555         /**
 556          * Get the values passed in the query string.
 557          * No transformation is performed on the values.
 558          *
 559          * @return Array
 560          */
 561          public function getQueryValues() {
 562                 return $_GET;
 563          }
 564
 565         /**
 566          * Get the HTTP method used for this request.
 567          *
 568          * @return String
 569          */
 570         public function getMethod() {
 571                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 572         }
 573
 574         /**
 575          * Returns true if the present request was reached by a POST operation,
 576          * false otherwise (GET, HEAD, or command-line).
 577          *
 578          * Note that values retrieved by the object may come from the
 579          * GET URL etc even on a POST request.
 580          *
 581          * @return Boolean
 582          */
 583         public function wasPosted() {
 584                 return $this->getMethod() == 'POST';
 585         }
 586
 587         /**
 588          * Returns true if there is a session cookie set.
 589          * This does not necessarily mean that the user is logged in!
 590          *
 591          * If you want to check for an open session, use session_id()
 592          * instead; that will also tell you if the session was opened
 593          * during the current request (in which case the cookie will
 594          * be sent back to the client at the end of the script run).
 595          *
 596          * @return Boolean
 597          */
 598         public function checkSessionCookie() {
 599                 return isset( $_COOKIE[ session_name() ] );
 600         }
 601
 602         /**
 603          * Get a cookie from the $_COOKIE jar
 604          *
 605          * @param $key String: the name of the cookie
 606          * @param $prefix String: a prefix to use for the cookie name, if not $wgCookiePrefix
 607          * @param $default Mixed: what to return if the value isn't found
 608          * @return Mixed: cookie value or $default if the cookie not set
 609          */
 610         public function getCookie( $key, $prefix = null, $default = null ) {
 611                 if( $prefix === null ) {
 612                         global $wgCookiePrefix;
 613                         $prefix = $wgCookiePrefix;
 614                 }
 615                 return $this->getGPCVal( $_COOKIE, $prefix . $key , $default );
 616         }
 617
 618         /**
 619          * Return the path and query string portion of the request URI.
 620          * This will be suitable for use as a relative link in HTML output.
 621          *
 622          * @throws MWException
 623          * @return String
 624          */
 625         public function getRequestURL() {
 626                 if( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 627                         $base = $_SERVER['REQUEST_URI'];
 628                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] ) && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] ) ) {
 629                         // Probably IIS; doesn't set REQUEST_URI
 630                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 631                 } elseif( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 632                         $base = $_SERVER['SCRIPT_NAME'];
 633                         if( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 634                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 635                         }
 636                 } else {
 637                         // This shouldn't happen!
 638                         throw new MWException( "Web server doesn't provide either " .
 639                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 640                                 "of your web server configuration to http://bugzilla.wikimedia.org/" );
 641                 }
 642                 // User-agents should not send a fragment with the URI, but
 643                 // if they do, and the web server passes it on to us, we
 644                 // need to strip it or we get false-positive redirect loops
 645                 // or weird output URLs
 646                 $hash = strpos( $base, '#' );
 647                 if( $hash !== false ) {
 648                         $base = substr( $base, 0, $hash );
 649                 }
 650                 if( $base[0] == '/' ) {
 651                         return $base;
 652                 } else {
 653                         // We may get paths with a host prepended; strip it.
 654                         return preg_replace( '!^[^:]+://[^/]+/!', '/', $base );
 655                 }
 656         }
 657
 658         /**
 659          * Return the request URI with the canonical service and hostname, path,
 660          * and query string. This will be suitable for use as an absolute link
 661          * in HTML or other output.
 662          *
 663          * If $wgServer is protocol-relative, this will return a fully
 664          * qualified URL with the protocol that was used for this request.
 665          *
 666          * @return String
 667          */
 668         public function getFullRequestURL() {
 669                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 670         }
 671
 672         /**
 673          * Take an arbitrary query and rewrite the present URL to include it
 674          * @param $query String: query string fragment; do not include initial '?'
 675          *
 676          * @return String
 677          */
 678         public function appendQuery( $query ) {
 679                 return $this->appendQueryArray( wfCgiToArray( $query ) );
 680         }
 681
 682         /**
 683          * HTML-safe version of appendQuery().
 684          * @deprecated: Deprecated in 1.20, warnings in 1.21, remove in 1.22.
 685          *
 686          * @param $query String: query string fragment; do not include initial '?'
 687          * @return String
 688          */
 689         public function escapeAppendQuery( $query ) {
 690                 return htmlspecialchars( $this->appendQuery( $query ) );
 691         }
 692
 693         /**
 694          * @param $key
 695          * @param $value
 696          * @param $onlyquery bool
 697          * @return String
 698          */
 699         public function appendQueryValue( $key, $value, $onlyquery = false ) {
 700                 return $this->appendQueryArray( array( $key => $value ), $onlyquery );
 701         }
 702
 703         /**
 704          * Appends or replaces value of query variables.
 705          *
 706          * @param $array Array of values to replace/add to query
 707          * @param $onlyquery Bool: whether to only return the query string and not
 708          *                   the complete URL
 709          * @return String
 710          */
 711         public function appendQueryArray( $array, $onlyquery = false ) {
 712                 global $wgTitle;
 713                 $newquery = $this->getQueryValues();
 714                 unset( $newquery['title'] );
 715                 $newquery = array_merge( $newquery, $array );
 716                 $query = wfArrayToCGI( $newquery );
 717                 return $onlyquery ? $query : $wgTitle->getLocalURL( $query );
 718         }
 719
 720         /**
 721          * Check for limit and offset parameters on the input, and return sensible
 722          * defaults if not given. The limit must be positive and is capped at 5000.
 723          * Offset must be positive but is not capped.
 724          *
 725          * @param $deflimit Integer: limit to use if no input and the user hasn't set the option.
 726          * @param $optionname String: to specify an option other than rclimit to pull from.
 727          * @return array first element is limit, second is offset
 728          */
 729         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 730                 global $wgUser;
 731
 732                 $limit = $this->getInt( 'limit', 0 );
 733                 if( $limit < 0 ) {
 734                         $limit = 0;
 735                 }
 736                 if( ( $limit == 0 ) && ( $optionname != '' ) ) {
 737                         $limit = (int)$wgUser->getOption( $optionname );
 738                 }
 739                 if( $limit <= 0 ) {
 740                         $limit = $deflimit;
 741                 }
 742                 if( $limit > 5000 ) {
 743                         $limit = 5000; # We have *some* limits...
 744                 }
 745
 746                 $offset = $this->getInt( 'offset', 0 );
 747                 if( $offset < 0 ) {
 748                         $offset = 0;
 749                 }
 750
 751                 return array( $limit, $offset );
 752         }
 753
 754         /**
 755          * Return the path to the temporary file where PHP has stored the upload.
 756          *
 757          * @param $key String:
 758          * @return string or NULL if no such file.
 759          */
 760         public function getFileTempname( $key ) {
 761                 $file = new WebRequestUpload( $this, $key );
 762                 return $file->getTempName();
 763         }
 764
 765         /**
 766          * Return the size of the upload, or 0.
 767          *
 768          * @deprecated since 1.17
 769          * @param $key String:
 770          * @return integer
 771          */
 772         public function getFileSize( $key ) {
 773                 wfDeprecated( __METHOD__, '1.17' );
 774                 $file = new WebRequestUpload( $this, $key );
 775                 return $file->getSize();
 776         }
 777
 778         /**
 779          * Return the upload error or 0
 780          *
 781          * @param $key String:
 782          * @return integer
 783          */
 784         public function getUploadError( $key ) {
 785                 $file = new WebRequestUpload( $this, $key );
 786                 return $file->getError();
 787         }
 788
 789         /**
 790          * Return the original filename of the uploaded file, as reported by
 791          * the submitting user agent. HTML-style character entities are
 792          * interpreted and normalized to Unicode normalization form C, in part
 793          * to deal with weird input from Safari with non-ASCII filenames.
 794          *
 795          * Other than this the name is not verified for being a safe filename.
 796          *
 797          * @param $key String:
 798          * @return string or NULL if no such file.
 799          */
 800         public function getFileName( $key ) {
 801                 $file = new WebRequestUpload( $this, $key );
 802                 return $file->getName();
 803         }
 804
 805         /**
 806          * Return a WebRequestUpload object corresponding to the key
 807          *
 808          * @param $key string
 809          * @return WebRequestUpload
 810          */
 811         public function getUpload( $key ) {
 812                 return new WebRequestUpload( $this, $key );
 813         }
 814
 815         /**
 816          * Return a handle to WebResponse style object, for setting cookies,
 817          * headers and other stuff, for Request being worked on.
 818          *
 819          * @return WebResponse
 820          */
 821         public function response() {
 822                 /* Lazy initialization of response object for this request */
 823                 if ( !is_object( $this->response ) ) {
 824                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 825                         $this->response = new $class();
 826                 }
 827                 return $this->response;
 828         }
 829
 830         /**
 831          * Initialise the header list
 832          */
 833         private function initHeaders() {
 834                 if ( count( $this->headers ) ) {
 835                         return;
 836                 }
 837
 838                 if ( function_exists( 'apache_request_headers' ) ) {
 839                         foreach ( apache_request_headers() as $tempName => $tempValue ) {
 840                                 $this->headers[ strtoupper( $tempName ) ] = $tempValue;
 841                         }
 842                 } else {
 843                         foreach ( $_SERVER as $name => $value ) {
 844                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 845                                         $name = str_replace( '_', '-',  substr( $name, 5 ) );
 846                                         $this->headers[$name] = $value;
 847                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 848                                         $this->headers['CONTENT-LENGTH'] = $value;
 849                                 }
 850                         }
 851                 }
 852         }
 853
 854         /**
 855          * Get an array containing all request headers
 856          *
 857          * @return Array mapping header name to its value
 858          */
 859         public function getAllHeaders() {
 860                 $this->initHeaders();
 861                 return $this->headers;
 862         }
 863
 864         /**
 865          * Get a request header, or false if it isn't set
 866          * @param $name String: case-insensitive header name
 867          *
 868          * @return string|bool False on failure
 869          */
 870         public function getHeader( $name ) {
 871                 $this->initHeaders();
 872                 $name = strtoupper( $name );
 873                 if ( isset( $this->headers[$name] ) ) {
 874                         return $this->headers[$name];
 875                 } else {
 876                         return false;
 877                 }
 878         }
 879
 880         /**
 881          * Get data from $_SESSION
 882          *
 883          * @param $key String: name of key in $_SESSION
 884          * @return Mixed
 885          */
 886         public function getSessionData( $key ) {
 887                 if( !isset( $_SESSION[$key] ) ) {
 888                         return null;
 889                 }
 890                 return $_SESSION[$key];
 891         }
 892
 893         /**
 894          * Set session data
 895          *
 896          * @param $key String: name of key in $_SESSION
 897          * @param $data Mixed
 898          */
 899         public function setSessionData( $key, $data ) {
 900                 $_SESSION[$key] = $data;
 901         }
 902
 903         /**
 904          * Check if Internet Explorer will detect an incorrect cache extension in
 905          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 906          * message or redirect to a safer URL. Returns true if the URL is OK, and
 907          * false if an error message has been shown and the request should be aborted.
 908          *
 909          * @param $extWhitelist array
 910          * @throws HttpError
 911          * @return bool
 912          */
 913         public function checkUrlExtension( $extWhitelist = array() ) {
 914                 global $wgScriptExtension;
 915                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 916                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 917                         if ( !$this->wasPosted() ) {
 918                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 919                                         $this->getFullRequestURL(), $extWhitelist );
 920                                 if ( $newUrl !== false ) {
 921                                         $this->doSecurityRedirect( $newUrl );
 922                                         return false;
 923                                 }
 924                         }
 925                         throw new HttpError( 403,
 926                                 'Invalid file extension found in the path info or query string.' );
 927                 }
 928                 return true;
 929         }
 930
 931         /**
 932          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 933          * IE 6. Returns true if it was successful, false otherwise.
 934          *
 935          * @param $url string
 936          * @return bool
 937          */
 938         protected function doSecurityRedirect( $url ) {
 939                 header( 'Location: ' . $url );
 940                 header( 'Content-Type: text/html' );
 941                 $encUrl = htmlspecialchars( $url );
 942                 echo <<<HTML
 943 <html>
 944 <head>
 945 <title>Security redirect</title>
 946 </head>
 947 <body>
 948 <h1>Security redirect</h1>
 949 <p>
 950 We can't serve non-HTML content from the URL you have requested, because
 951 Internet Explorer would interpret it as an incorrect and potentially dangerous
 952 content type.</p>
 953 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the URL you have requested, except that
 954 "&amp;*" is appended. This prevents Internet Explorer from seeing a bogus file
 955 extension.
 956 </p>
 957 </body>
 958 </html>
 959 HTML;
 960                 echo "\n";
 961                 return true;
 962         }
 963
 964         /**
 965          * Returns true if the PATH_INFO ends with an extension other than a script
 966          * extension. This could confuse IE for scripts that send arbitrary data which
 967          * is not HTML but may be detected as such.
 968          *
 969          * Various past attempts to use the URL to make this check have generally
 970          * run up against the fact that CGI does not provide a standard method to
 971          * determine the URL. PATH_INFO may be mangled (e.g. if cgi.fix_pathinfo=0),
 972          * but only by prefixing it with the script name and maybe some other stuff,
 973          * the extension is not mangled. So this should be a reasonably portable
 974          * way to perform this security check.
 975          *
 976          * Also checks for anything that looks like a file extension at the end of
 977          * QUERY_STRING, since IE 6 and earlier will use this to get the file type
 978          * if there was no dot before the question mark (bug 28235).
 979          *
 980          * @deprecated Use checkUrlExtension().
 981          *
 982          * @param $extWhitelist array
 983          *
 984          * @return bool
 985          */
 986         public function isPathInfoBad( $extWhitelist = array() ) {
 987                 wfDeprecated( __METHOD__, '1.17' );
 988                 global $wgScriptExtension;
 989                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 990                 return IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist );
 991         }
 992
 993         /**
 994          * Parse the Accept-Language header sent by the client into an array
 995          * @return array array( languageCode => q-value ) sorted by q-value in descending order then
 996          *                                                appearing time in the header in ascending order.
 997          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 998          * This is aligned with rfc2616 section 14.4
 999          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1000          */
1001         public function getAcceptLang() {
1002                 // Modified version of code found at http://www.thefutureoftheweb.com/blog/use-accept-language-header
1003                 $acceptLang = $this->getHeader( 'Accept-Language' );
1004                 if ( !$acceptLang ) {
1005                         return array();
1006                 }
1007
1008                 // Return the language codes in lower case
1009                 $acceptLang = strtolower( $acceptLang );
1010
1011                 // Break up string into pieces (languages and q factors)
1012                 $lang_parse = null;
1013                 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})?)?)?/',
1014                         $acceptLang, $lang_parse );
1015
1016                 if ( !count( $lang_parse[1] ) ) {
1017                         return array();
1018                 }
1019
1020                 $langcodes = $lang_parse[1];
1021                 $qvalues = $lang_parse[4];
1022                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1023
1024                 // Set default q factor to 1
1025                 foreach ( $indices as $index ) {
1026                         if ( $qvalues[$index] === '' ) {
1027                                 $qvalues[$index] = 1;
1028                         } elseif ( $qvalues[$index] == 0 ) {
1029                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1030                         }
1031                 }
1032
1033                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1034                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1035
1036                 // Create a list like "en" => 0.8
1037                 $langs = array_combine( $langcodes, $qvalues );
1038
1039                 return $langs;
1040         }
1041
1042         /**
1043          * Fetch the raw IP from the request
1044          *
1045          * @since 1.19
1046          *
1047          * @return String
1048          */
1049         protected function getRawIP() {
1050                 if ( isset( $_SERVER['REMOTE_ADDR'] ) ) {
1051                         return IP::canonicalize( $_SERVER['REMOTE_ADDR'] );
1052                 } else {
1053                         return null;
1054                 }
1055         }
1056
1057         /**
1058          * Work out the IP address based on various globals
1059          * For trusted proxies, use the XFF client IP (first of the chain)
1060          *
1061          * @since 1.19
1062          *
1063          * @throws MWException
1064          * @return string
1065          */
1066         public function getIP() {
1067                 global $wgUsePrivateIPs;
1068
1069                 # Return cached result
1070                 if ( $this->ip !== null ) {
1071                         return $this->ip;
1072                 }
1073
1074                 # collect the originating ips
1075                 $ip = $this->getRawIP();
1076
1077                 # Append XFF
1078                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1079                 if ( $forwardedFor !== false ) {
1080                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1081                         $ipchain = array_reverse( $ipchain );
1082                         if ( $ip ) {
1083                                 array_unshift( $ipchain, $ip );
1084                         }
1085
1086                         # Step through XFF list and find the last address in the list which is a trusted server
1087                         # Set $ip to the IP address given by that trusted server, unless the address is not sensible (e.g. private)
1088                         foreach ( $ipchain as $i => $curIP ) {
1089                                 $curIP = IP::canonicalize( $curIP );
1090                                 if ( wfIsTrustedProxy( $curIP ) ) {
1091                                         if ( isset( $ipchain[$i + 1] ) ) {
1092                                                 if ( $wgUsePrivateIPs || IP::isPublic( $ipchain[$i + 1 ] ) ) {
1093                                                         $ip = $ipchain[$i + 1];
1094                                                 }
1095                                         }
1096                                 } else {
1097                                         break;
1098                                 }
1099                         }
1100                 }
1101
1102                 # Allow extensions to improve our guess
1103                 wfRunHooks( 'GetIP', array( &$ip ) );
1104
1105                 if ( !$ip ) {
1106                         throw new MWException( "Unable to determine IP" );
1107                 }
1108
1109                 wfDebug( "IP: $ip\n" );
1110                 $this->ip = $ip;
1111                 return $ip;
1112         }
1113 }
1114
1115 /**
1116  * Object to access the $_FILES array
1117  */
1118 class WebRequestUpload {
1119         protected $request;
1120         protected $doesExist;
1121         protected $fileInfo;
1122
1123         /**
1124          * Constructor. Should only be called by WebRequest
1125          *
1126          * @param $request WebRequest The associated request
1127          * @param $key string Key in $_FILES array (name of form field)
1128          */
1129         public function __construct( $request, $key ) {
1130                 $this->request = $request;
1131                 $this->doesExist = isset( $_FILES[$key] );
1132                 if ( $this->doesExist ) {
1133                         $this->fileInfo = $_FILES[$key];
1134                 }
1135         }
1136
1137         /**
1138          * Return whether a file with this name was uploaded.
1139          *
1140          * @return bool
1141          */
1142         public function exists() {
1143                 return $this->doesExist;
1144         }
1145
1146         /**
1147          * Return the original filename of the uploaded file
1148          *
1149          * @return mixed Filename or null if non-existent
1150          */
1151         public function getName() {
1152                 if ( !$this->exists() ) {
1153                         return null;
1154                 }
1155
1156                 global $wgContLang;
1157                 $name = $this->fileInfo['name'];
1158
1159                 # Safari sends filenames in HTML-encoded Unicode form D...
1160                 # Horrid and evil! Let's try to make some kind of sense of it.
1161                 $name = Sanitizer::decodeCharReferences( $name );
1162                 $name = $wgContLang->normalize( $name );
1163                 wfDebug( __METHOD__ . ": {$this->fileInfo['name']} normalized to '$name'\n" );
1164                 return $name;
1165         }
1166
1167         /**
1168          * Return the file size of the uploaded file
1169          *
1170          * @return int File size or zero if non-existent
1171          */
1172         public function getSize() {
1173                 if ( !$this->exists() ) {
1174                         return 0;
1175                 }
1176
1177                 return $this->fileInfo['size'];
1178         }
1179
1180         /**
1181          * Return the path to the temporary file
1182          *
1183          * @return mixed Path or null if non-existent
1184          */
1185         public function getTempName() {
1186                 if ( !$this->exists() ) {
1187                         return null;
1188                 }
1189
1190                 return $this->fileInfo['tmp_name'];
1191         }
1192
1193         /**
1194          * Return the upload error. See link for explanation
1195          * http://www.php.net/manual/en/features.file-upload.errors.php
1196          *
1197          * @return int One of the UPLOAD_ constants, 0 if non-existent
1198          */
1199         public function getError() {
1200                 if ( !$this->exists() ) {
1201                         return 0; # UPLOAD_ERR_OK
1202                 }
1203
1204                 return $this->fileInfo['error'];
1205         }
1206
1207         /**
1208          * Returns whether this upload failed because of overflow of a maximum set
1209          * in php.ini
1210          *
1211          * @return bool
1212          */
1213         public function isIniSizeOverflow() {
1214                 if ( $this->getError() == UPLOAD_ERR_INI_SIZE ) {
1215                         # PHP indicated that upload_max_filesize is exceeded
1216                         return true;
1217                 }
1218
1219                 $contentLength = $this->request->getHeader( 'CONTENT_LENGTH' );
1220                 if ( $contentLength > wfShorthandToInteger( ini_get( 'post_max_size' ) ) ) {
1221                         # post_max_size is exceeded
1222                         return true;
1223                 }
1224
1225                 return false;
1226         }
1227 }
1228
1229 /**
1230  * WebRequest clone which takes values from a provided array.
1231  *
1232  * @ingroup HTTP
1233  */
1234 class FauxRequest extends WebRequest {
1235         private $wasPosted = false;
1236         private $session = array();
1237
1238         /**
1239          * @param $data Array of *non*-urlencoded key => value pairs, the
1240          *   fake GET/POST values
1241          * @param $wasPosted Bool: whether to treat the data as POST
1242          * @param $session Mixed: session array or null
1243          * @throws MWException
1244          */
1245         public function __construct( $data = array(), $wasPosted = false, $session = null ) {
1246                 if( is_array( $data ) ) {
1247                         $this->data = $data;
1248                 } else {
1249                         throw new MWException( "FauxRequest() got bogus data" );
1250                 }
1251                 $this->wasPosted = $wasPosted;
1252                 if( $session )
1253                         $this->session = $session;
1254         }
1255
1256         /**
1257          * @param $method string
1258          * @throws MWException
1259          */
1260         private function notImplemented( $method ) {
1261                 throw new MWException( "{$method}() not implemented" );
1262         }
1263
1264         /**
1265          * @param $name string
1266          * @param $default string
1267          * @return string
1268          */
1269         public function getText( $name, $default = '' ) {
1270                 # Override; don't recode since we're using internal data
1271                 return (string)$this->getVal( $name, $default );
1272         }
1273
1274         /**
1275          * @return Array
1276          */
1277         public function getValues() {
1278                 return $this->data;
1279         }
1280
1281         /**
1282          * @return array
1283          */
1284         public function getQueryValues() {
1285                 if ( $this->wasPosted ) {
1286                         return array();
1287                 } else {
1288                         return $this->data;
1289                 }
1290         }
1291
1292         public function getMethod() {
1293                 return $this->wasPosted ? 'POST' : 'GET';
1294         }
1295
1296         /**
1297          * @return bool
1298          */
1299         public function wasPosted() {
1300                 return $this->wasPosted;
1301         }
1302
1303         public function checkSessionCookie() {
1304                 return false;
1305         }
1306
1307         public function getRequestURL() {
1308                 $this->notImplemented( __METHOD__ );
1309         }
1310
1311         /**
1312          * @param $name
1313          * @return bool|string
1314          */
1315         public function getHeader( $name ) {
1316                 return isset( $this->headers[$name] ) ? $this->headers[$name] : false;
1317         }
1318
1319         /**
1320          * @param $name string
1321          * @param $val string
1322          */
1323         public function setHeader( $name, $val ) {
1324                 $this->headers[$name] = $val;
1325         }
1326
1327         /**
1328          * @param $key
1329          * @return mixed
1330          */
1331         public function getSessionData( $key ) {
1332                 if( isset( $this->session[$key] ) )
1333                         return $this->session[$key];
1334         }
1335
1336         /**
1337          * @param $key
1338          * @param $data
1339          */
1340         public function setSessionData( $key, $data ) {
1341                 $this->session[$key] = $data;
1342         }
1343
1344         /**
1345          * @return array|Mixed|null
1346          */
1347         public function getSessionArray() {
1348                 return $this->session;
1349         }
1350
1351         /**
1352          * @param array $extWhitelist
1353          * @return bool
1354          */
1355         public function isPathInfoBad( $extWhitelist = array() ) {
1356                 return false;
1357         }
1358
1359         /**
1360          * @param array $extWhitelist
1361          * @return bool
1362          */
1363         public function checkUrlExtension( $extWhitelist = array() ) {
1364                 return true;
1365         }
1366
1367         /**
1368          * @return string
1369          */
1370         protected function getRawIP() {
1371                 return '127.0.0.1';
1372         }
1373 }
1374
1375 /**
1376  * Similar to FauxRequest, but only fakes URL parameters and method
1377  * (POST or GET) and use the base request for the remaining stuff
1378  * (cookies, session and headers).
1379  *
1380  * @ingroup HTTP
1381  * @since 1.19
1382  */
1383 class DerivativeRequest extends FauxRequest {
1384         private $base;
1385
1386         public function __construct( WebRequest $base, $data, $wasPosted = false ) {
1387                 $this->base = $base;
1388                 parent::__construct( $data, $wasPosted );
1389         }
1390
1391         public function getCookie( $key, $prefix = null, $default = null ) {
1392                 return $this->base->getCookie( $key, $prefix, $default );
1393         }
1394
1395         public function checkSessionCookie() {
1396                 return $this->base->checkSessionCookie();
1397         }
1398
1399         public function getHeader( $name ) {
1400                 return $this->base->getHeader( $name );
1401         }
1402
1403         public function getAllHeaders() {
1404                 return $this->base->getAllHeaders();
1405         }
1406
1407         public function getSessionData( $key ) {
1408                 return $this->base->getSessionData( $key );
1409         }
1410
1411         public function setSessionData( $key, $data ) {
1412                 $this->base->setSessionData( $key, $data );
1413         }
1414
1415         public function getAcceptLang() {
1416                 return $this->base->getAcceptLang();
1417         }
1418
1419         public function getIP() {
1420                 return $this->base->getIP();
1421         }
1422 }