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