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