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