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