includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasssty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * The WebRequest class encapsulates getting at data passed in the
  28  * URL or via a POSTed form, handling remove of "magic quotes" slashes,
  29  * stripping illegal input characters and normalizing Unicode sequences.
  30  *
  31  * Usually this is used via a global singleton, $wgRequest. You should
  32  * not create a second WebRequest object; make a FauxRequest object if
  33  * you want to pass arbitrary data to some function in place of the web
  34  * input.
  35  *
  36  * @ingroup HTTP
  37  */
  38 class WebRequest {
  39         protected $data, $headers = array();
  40
  41         /**
  42          * Lazy-init response object
  43          * @var WebResponse
  44          */
  45         private $response;
  46
  47         /**
  48          * Cached client IP address
  49          * @var String
  50          */
  51         private $ip;
  52
  53         public function __construct() {
  54                 /// @todo FIXME: This preemptive de-quoting can interfere with other web libraries
  55                 ///        and increases our memory footprint. It would be cleaner to do on
  56                 ///        demand; but currently we have no wrapper for $_SERVER etc.
  57                 $this->checkMagicQuotes();
  58
  59                 // POST overrides GET data
  60                 // We don't use $_REQUEST here to avoid interference from cookies...
  61                 $this->data = $_POST + $_GET;
  62         }
  63
  64         /**
  65          * Extract the PATH_INFO variable even when it isn't a reasonable
  66          * value. On some large webhosts, PATH_INFO includes the script
  67          * path as well as everything after it.
  68          *
  69          * @param $want string: If this is not 'all', then the function
  70          * will return an empty array if it determines that the URL is
  71          * inside a rewrite path.
  72          *
  73          * @return Array: 'title' key is the title of the article.
  74          */
  75         static public function getPathInfo( $want = 'all' ) {
  76                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
  77                 // And also by Apache 2.x, double slashes are converted to single slashes.
  78                 // So we will use REQUEST_URI if possible.
  79                 $matches = array();
  80                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
  81                         // Slurp out the path portion to examine...
  82                         $url = $_SERVER['REQUEST_URI'];
  83                         if ( !preg_match( '!^https?://!', $url ) ) {
  84                                 $url = 'http://unused' . $url;
  85                         }
  86                         $a = parse_url( $url );
  87                         if( $a ) {
  88                                 $path = isset( $a['path'] ) ? $a['path'] : '';
  89
  90                                 global $wgScript;
  91                                 if( $path == $wgScript && $want !== 'all' ) {
  92                                         // Script inside a rewrite path?
  93                                         // Abort to keep from breaking...
  94                                         return $matches;
  95                                 }
  96                                 // Raw PATH_INFO style
  97                                 $matches = self::extractTitle( $path, "$wgScript/$1" );
  98
  99                                 if( !$matches
 100                                         && isset( $_SERVER['SCRIPT_NAME'] )
 101                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] ) )
 102                                 {
 103                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 104                                         # But we do have some other .php files such as img_auth.php
 105                                         # Don't let root article paths clober the parsing for them
 106                                         $matches = self::extractTitle( $path, $_SERVER['SCRIPT_NAME'] . "/$1" );
 107                                 }
 108
 109                                 global $wgArticlePath;
 110                                 if( !$matches && $wgArticlePath ) {
 111                                         $matches = self::extractTitle( $path, $wgArticlePath );
 112                                 }
 113
 114                                 global $wgActionPaths;
 115                                 if( !$matches && $wgActionPaths ) {
 116                                         $matches = self::extractTitle( $path, $wgActionPaths, 'action' );
 117                                 }
 118
 119                                 global $wgVariantArticlePath, $wgContLang;
 120                                 if( !$matches && $wgVariantArticlePath ) {
 121                                         $variantPaths = array();
 122                                         foreach( $wgContLang->getVariants() as $variant ) {
 123                                                 $variantPaths[$variant] =
 124                                                         str_replace( '$2', $variant, $wgVariantArticlePath );
 125                                         }
 126                                         $matches = self::extractTitle( $path, $variantPaths, 'variant' );
 127                                 }
 128
 129                                 wfRunHooks( 'WebRequestGetPathInfoRequestURI', array( $path, &$matches ) );
 130                         }
 131                 } elseif ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 132                         // Mangled PATH_INFO
 133                         // http://bugs.php.net/bug.php?id=31892
 134                         // Also reported when ini_get('cgi.fix_pathinfo')==false
 135                         $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 136
 137                 } elseif ( isset( $_SERVER['PATH_INFO'] ) && ($_SERVER['PATH_INFO'] != '') ) {
 138                         // Regular old PATH_INFO yay
 139                         $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 140                 }
 141
 142                 return $matches;
 143         }
 144
 145         /**
 146          * Work out an appropriate URL prefix containing scheme and host, based on
 147          * information detected from $_SERVER
 148          *
 149          * @return string
 150          */
 151         public static function detectServer() {
 152                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 153
 154                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 155                 $host = 'localhost';
 156                 $port = $stdPort;
 157                 foreach ( $varNames as $varName ) {
 158                         if ( !isset( $_SERVER[$varName] ) ) {
 159                                 continue;
 160                         }
 161                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 162                         if ( !$parts ) {
 163                                 // Invalid, do not use
 164                                 continue;
 165                         }
 166                         $host = $parts[0];
 167                         if ( $parts[1] === false ) {
 168                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 169                                         $port = $_SERVER['SERVER_PORT'];
 170                                 } // else leave it as $stdPort
 171                         } else {
 172                                 $port = $parts[1];
 173                         }
 174                         break;
 175                 }
 176
 177                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 178         }
 179
 180         /**
 181          * @return array
 182          */
 183         public static function detectProtocolAndStdPort() {
 184                 return ( isset( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] == 'on' ) ? array( 'https', 443 ) : array( 'http', 80 );
 185         }
 186
 187         /**
 188          * @return string
 189          */
 190         public static function detectProtocol() {
 191                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 192                 return $proto;
 193         }
 194
 195         /**
 196          * Check for title, action, and/or variant data in the URL
 197          * and interpolate it into the GET variables.
 198          * This should only be run after $wgContLang is available,
 199          * as we may need the list of language variants to determine
 200          * available variant URLs.
 201          */
 202         public function interpolateTitle() {
 203                 global $wgUsePathInfo;
 204
 205                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 206                 if ( defined( 'MW_API' ) ) {
 207                         return;
 208                 }
 209
 210                 if ( $wgUsePathInfo ) {
 211                         $matches = self::getPathInfo( 'title' );
 212                         foreach( $matches as $key => $val) {
 213                                 $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 214                         }
 215                 }
 216         }
 217
 218         /**
 219          * URL rewriting function; tries to extract page title and,
 220          * optionally, one other fixed parameter value from a URL path.
 221          *
 222          * @param $path string: the URL path given from the client
 223          * @param $bases array: one or more URLs, optionally with $1 at the end
 224          * @param $key string: if provided, the matching key in $bases will be
 225          *             passed on as the value of this URL parameter
 226          * @return array of URL variables to interpolate; empty if no match
 227          */
 228         static function extractTitle( $path, $bases, $key = false ) {
 229                 foreach( (array)$bases as $keyValue => $base ) {
 230                         // Find the part after $wgArticlePath
 231                         $base = str_replace( '$1', '', $base );
 232                         $baseLen = strlen( $base );
 233                         if( substr( $path, 0, $baseLen ) == $base ) {
 234                                 $raw = substr( $path, $baseLen );
 235                                 if( $raw !== '' ) {
 236                                         $matches = array( 'title' => rawurldecode( $raw ) );
 237                                         if( $key ) {
 238                                                 $matches[$key] = $keyValue;
 239                                         }
 240                                         return $matches;
 241                                 }
 242                         }
 243                 }
 244                 return array();
 245         }
 246
 247         /**
 248          * Recursively strips slashes from the given array;
 249          * used for undoing the evil that is magic_quotes_gpc.
 250          *
 251          * @param $arr array: will be modified
 252          * @param $topLevel bool Specifies if the array passed is from the top
 253          * level of the source. In PHP5 magic_quotes only escapes the first level
 254          * of keys that belong to an array.
 255          * @return array the original array
 256          * @see http://www.php.net/manual/en/function.get-magic-quotes-gpc.php#49612
 257          */
 258         private function &fix_magic_quotes( &$arr, $topLevel = true ) {
 259                 $clean = array();
 260                 foreach( $arr as $key => $val ) {
 261                         if( is_array( $val ) ) {
 262                                 $cleanKey = $topLevel ? stripslashes( $key ) : $key;
 263                                 $clean[$cleanKey] = $this->fix_magic_quotes( $arr[$key], false );
 264                         } else {
 265                                 $cleanKey = stripslashes( $key );
 266                                 $clean[$cleanKey] = stripslashes( $val );
 267                         }
 268                 }
 269                 $arr = $clean;
 270                 return $arr;
 271         }
 272
 273         /**
 274          * If magic_quotes_gpc option is on, run the global arrays
 275          * through fix_magic_quotes to strip out the stupid slashes.
 276          * WARNING: This should only be done once! Running a second
 277          * time could damage the values.
 278          */
 279         private function checkMagicQuotes() {
 280                 $mustFixQuotes = function_exists( 'get_magic_quotes_gpc' )
 281                         && get_magic_quotes_gpc();
 282                 if( $mustFixQuotes ) {
 283                         $this->fix_magic_quotes( $_COOKIE );
 284                         $this->fix_magic_quotes( $_ENV );
 285                         $this->fix_magic_quotes( $_GET );
 286                         $this->fix_magic_quotes( $_POST );
 287                         $this->fix_magic_quotes( $_REQUEST );
 288                         $this->fix_magic_quotes( $_SERVER );
 289                 }
 290         }
 291
 292         /**
 293          * Recursively normalizes UTF-8 strings in the given array.
 294          *
 295          * @param $data string or array
 296          * @return cleaned-up version of the given
 297          * @private
 298          */
 299         function normalizeUnicode( $data ) {
 300                 if( is_array( $data ) ) {
 301                         foreach( $data as $key => $val ) {
 302                                 $data[$key] = $this->normalizeUnicode( $val );
 303                         }
 304                 } else {
 305                         global $wgContLang;
 306                         $data = isset( $wgContLang ) ? $wgContLang->normalize( $data ) : UtfNormal::cleanUp( $data );
 307                 }
 308                 return $data;
 309         }
 310
 311         /**
 312          * Fetch a value from the given array or return $default if it's not set.
 313          *
 314          * @param $arr Array
 315          * @param $name String
 316          * @param $default Mixed
 317          * @return mixed
 318          */
 319         private function getGPCVal( $arr, $name, $default ) {
 320                 # PHP is so nice to not touch input data, except sometimes:
 321                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 322                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 323                 $name = strtr( $name, '.', '_' );
 324                 if( isset( $arr[$name] ) ) {
 325                         global $wgContLang;
 326                         $data = $arr[$name];
 327                         if( isset( $_GET[$name] ) && !is_array( $data ) ) {
 328                                 # Check for alternate/legacy character encoding.
 329                                 if( isset( $wgContLang ) ) {
 330                                         $data = $wgContLang->checkTitleEncoding( $data );
 331                                 }
 332                         }
 333                         $data = $this->normalizeUnicode( $data );
 334                         return $data;
 335                 } else {
 336                         taint( $default );
 337                         return $default;
 338                 }
 339         }
 340
 341         /**
 342          * Fetch a scalar from the input or return $default if it's not set.
 343          * Returns a string. Arrays are discarded. Useful for
 344          * non-freeform text inputs (e.g. predefined internal text keys
 345          * selected by a drop-down menu). For freeform input, see getText().
 346          *
 347          * @param $name String
 348          * @param $default String: optional default (or NULL)
 349          * @return String
 350          */
 351         public function getVal( $name, $default = null ) {
 352                 $val = $this->getGPCVal( $this->data, $name, $default );
 353                 if( is_array( $val ) ) {
 354                         $val = $default;
 355                 }
 356                 if( is_null( $val ) ) {
 357                         return $val;
 358                 } else {
 359                         return (string)$val;
 360                 }
 361         }
 362
 363         /**
 364          * Set an arbitrary value into our get/post data.
 365          *
 366          * @param $key String: key name to use
 367          * @param $value Mixed: value to set
 368          * @return Mixed: old value if one was present, null otherwise
 369          */
 370         public function setVal( $key, $value ) {
 371                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 372                 $this->data[$key] = $value;
 373                 return $ret;
 374         }
 375
 376         /**
 377          * Fetch an array from the input or return $default if it's not set.
 378          * If source was scalar, will return an array with a single element.
 379          * If no source and no default, returns NULL.
 380          *
 381          * @param $name String
 382          * @param $default Array: optional default (or NULL)
 383          * @return Array
 384          */
 385         public function getArray( $name, $default = null ) {
 386                 $val = $this->getGPCVal( $this->data, $name, $default );
 387                 if( is_null( $val ) ) {
 388                         return null;
 389                 } else {
 390                         return (array)$val;
 391                 }
 392         }
 393
 394         /**
 395          * Fetch an array of integers, or return $default if it's not set.
 396          * If source was scalar, will return an array with a single element.
 397          * If no source and no default, returns NULL.
 398          * If an array is returned, contents are guaranteed to be integers.
 399          *
 400          * @param $name String
 401          * @param $default Array: option default (or NULL)
 402          * @return Array of ints
 403          */
 404         public function getIntArray( $name, $default = null ) {
 405                 $val = $this->getArray( $name, $default );
 406                 if( is_array( $val ) ) {
 407                         $val = array_map( 'intval', $val );
 408                 }
 409                 return $val;
 410         }
 411
 412         /**
 413          * Fetch an integer value from the input or return $default if not set.
 414          * Guaranteed to return an integer; non-numeric input will typically
 415          * return 0.
 416          *
 417          * @param $name String
 418          * @param $default Integer
 419          * @return Integer
 420          */
 421         public function getInt( $name, $default = 0 ) {
 422                 return intval( $this->getVal( $name, $default ) );
 423         }
 424
 425         /**
 426          * Fetch an integer value from the input or return null if empty.
 427          * Guaranteed to return an integer or null; non-numeric input will
 428          * typically return null.
 429          *
 430          * @param $name String
 431          * @return Integer
 432          */
 433         public function getIntOrNull( $name ) {
 434                 $val = $this->getVal( $name );
 435                 return is_numeric( $val )
 436                         ? intval( $val )
 437                         : null;
 438         }
 439
 440         /**
 441          * Fetch a boolean value from the input or return $default if not set.
 442          * Guaranteed to return true or false, with normal PHP semantics for
 443          * boolean interpretation of strings.
 444          *
 445          * @param $name String
 446          * @param $default Boolean
 447          * @return Boolean
 448          */
 449         public function getBool( $name, $default = false ) {
 450                 return (bool)$this->getVal( $name, $default );
 451         }
 452
 453         /**
 454          * Fetch a boolean value from the input or return $default if not set.
 455          * Unlike getBool, the string "false" will result in boolean false, which is
 456          * useful when interpreting information sent from JavaScript.
 457          *
 458          * @param $name String
 459          * @param $default Boolean
 460          * @return Boolean
 461          */
 462         public function getFuzzyBool( $name, $default = false ) {
 463                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 464         }
 465
 466         /**
 467          * Return true if the named value is set in the input, whatever that
 468          * value is (even "0"). Return false if the named value is not set.
 469          * Example use is checking for the presence of check boxes in forms.
 470          *
 471          * @param $name String
 472          * @return Boolean
 473          */
 474         public function getCheck( $name ) {
 475                 # Checkboxes and buttons are only present when clicked
 476                 # Presence connotes truth, abscense false
 477                 $val = $this->getVal( $name, null );
 478                 return isset( $val );
 479         }
 480
 481         /**
 482          * Fetch a text string from the given array or return $default if it's not
 483          * set. Carriage returns are stripped from the text, and with some language
 484          * modules there is an input transliteration applied. This should generally
 485          * be used for form <textarea> and <input> fields. Used for user-supplied
 486          * freeform text input (for which input transformations may be required - e.g.
 487          * Esperanto x-coding).
 488          *
 489          * @param $name String
 490          * @param $default String: optional
 491          * @return String
 492          */
 493         public function getText( $name, $default = '' ) {
 494                 global $wgContLang;
 495                 $val = $this->getVal( $name, $default );
 496                 return str_replace( "\r\n", "\n",
 497                         $wgContLang->recodeInput( $val ) );
 498         }
 499
 500         /**
 501          * Extracts the given named values into an array.
 502          * If no arguments are given, returns all input values.
 503          * No transformation is performed on the values.
 504          *
 505          * @return array
 506          */
 507         public function getValues() {
 508                 $names = func_get_args();
 509                 if ( count( $names ) == 0 ) {
 510                         $names = array_keys( $this->data );
 511                 }
 512
 513                 $retVal = array();
 514                 foreach ( $names as $name ) {
 515                         $value = $this->getVal( $name );
 516                         if ( !is_null( $value ) ) {
 517                                 $retVal[$name] = $value;
 518                         }
 519                 }
 520                 return $retVal;
 521         }
 522
 523         /**
 524          * Returns the names of all input values excluding those in $exclude.
 525          *
 526          * @param $exclude Array
 527          * @return array
 528          */
 529         public function getValueNames( $exclude = array() ) {
 530                 return array_diff( array_keys( $this->getValues() ), $exclude );
 531         }
 532
 533         /**
 534          * Get the values passed in the query string.
 535          * No transformation is performed on the values.
 536          *
 537          * @return Array
 538          */
 539          public function getQueryValues() {
 540                 return $_GET;
 541          }
 542
 543         /**
 544          * Returns true if the present request was reached by a POST operation,
 545          * false otherwise (GET, HEAD, or command-line).
 546          *
 547          * Note that values retrieved by the object may come from the
 548          * GET URL etc even on a POST request.
 549          *
 550          * @return Boolean
 551          */
 552         public function wasPosted() {
 553                 return isset( $_SERVER['REQUEST_METHOD'] ) && $_SERVER['REQUEST_METHOD'] == 'POST';
 554         }
 555
 556         /**
 557          * Returns true if there is a session cookie set.
 558          * This does not necessarily mean that the user is logged in!
 559          *
 560          * If you want to check for an open session, use session_id()
 561          * instead; that will also tell you if the session was opened
 562          * during the current request (in which case the cookie will
 563          * be sent back to the client at the end of the script run).
 564          *
 565          * @return Boolean
 566          */
 567         public function checkSessionCookie() {
 568                 return isset( $_COOKIE[ session_name() ] );
 569         }
 570
 571         /**
 572          * Get a cookie from the $_COOKIE jar
 573          *
 574          * @param $key String: the name of the cookie
 575          * @param $prefix String: a prefix to use for the cookie name, if not $wgCookiePrefix
 576          * @param $default Mixed: what to return if the value isn't found
 577          * @return Mixed: cookie value or $default if the cookie not set
 578          */
 579         public function getCookie( $key, $prefix = null, $default = null ) {
 580                 if( $prefix === null ) {
 581                         global $wgCookiePrefix;
 582                         $prefix = $wgCookiePrefix;
 583                 }
 584                 return $this->getGPCVal( $_COOKIE, $prefix . $key , $default );
 585         }
 586
 587         /**
 588          * Return the path and query string portion of the request URI.
 589          * This will be suitable for use as a relative link in HTML output.
 590          *
 591          * @return String
 592          */
 593         public function getRequestURL() {
 594                 if( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 595                         $base = $_SERVER['REQUEST_URI'];
 596                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] ) && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] ) ) {
 597                         // Probably IIS; doesn't set REQUEST_URI
 598                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 599                 } elseif( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 600                         $base = $_SERVER['SCRIPT_NAME'];
 601                         if( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 602                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 603                         }
 604                 } else {
 605                         // This shouldn't happen!
 606                         throw new MWException( "Web server doesn't provide either " .
 607                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 608                                 "of your web server configuration to http://bugzilla.wikimedia.org/" );
 609                 }
 610                 // User-agents should not send a fragment with the URI, but
 611                 // if they do, and the web server passes it on to us, we
 612                 // need to strip it or we get false-positive redirect loops
 613                 // or weird output URLs
 614                 $hash = strpos( $base, '#' );
 615                 if( $hash !== false ) {
 616                         $base = substr( $base, 0, $hash );
 617                 }
 618                 if( $base[0] == '/' ) {
 619                         return $base;
 620                 } else {
 621                         // We may get paths with a host prepended; strip it.
 622                         return preg_replace( '!^[^:]+://[^/]+/!', '/', $base );
 623                 }
 624         }
 625
 626         /**
 627          * Return the request URI with the canonical service and hostname, path,
 628          * and query string. This will be suitable for use as an absolute link
 629          * in HTML or other output.
 630          *
 631          * If $wgServer is protocol-relative, this will return a fully
 632          * qualified URL with the protocol that was used for this request.
 633          *
 634          * @return String
 635          */
 636         public function getFullRequestURL() {
 637                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 638         }
 639
 640         /**
 641          * Take an arbitrary query and rewrite the present URL to include it
 642          * @param $query String: query string fragment; do not include initial '?'
 643          *
 644          * @return String
 645          */
 646         public function appendQuery( $query ) {
 647                 return $this->appendQueryArray( wfCgiToArray( $query ) );
 648         }
 649
 650         /**
 651          * HTML-safe version of appendQuery().
 652          *
 653          * @param $query String: query string fragment; do not include initial '?'
 654          * @return String
 655          */
 656         public function escapeAppendQuery( $query ) {
 657                 return htmlspecialchars( $this->appendQuery( $query ) );
 658         }
 659
 660         /**
 661          * @param $key
 662          * @param $value
 663          * @param $onlyquery bool
 664          * @return String
 665          */
 666         public function appendQueryValue( $key, $value, $onlyquery = false ) {
 667                 return $this->appendQueryArray( array( $key => $value ), $onlyquery );
 668         }
 669
 670         /**
 671          * Appends or replaces value of query variables.
 672          *
 673          * @param $array Array of values to replace/add to query
 674          * @param $onlyquery Bool: whether to only return the query string and not
 675          *                   the complete URL
 676          * @return String
 677          */
 678         public function appendQueryArray( $array, $onlyquery = false ) {
 679                 global $wgTitle;
 680                 $newquery = $this->getQueryValues();
 681                 unset( $newquery['title'] );
 682                 $newquery = array_merge( $newquery, $array );
 683                 $query = wfArrayToCGI( $newquery );
 684                 return $onlyquery ? $query : $wgTitle->getLocalURL( $query );
 685         }
 686
 687         /**
 688          * Check for limit and offset parameters on the input, and return sensible
 689          * defaults if not given. The limit must be positive and is capped at 5000.
 690          * Offset must be positive but is not capped.
 691          *
 692          * @param $deflimit Integer: limit to use if no input and the user hasn't set the option.
 693          * @param $optionname String: to specify an option other than rclimit to pull from.
 694          * @return array first element is limit, second is offset
 695          */
 696         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 697                 global $wgUser;
 698
 699                 $limit = $this->getInt( 'limit', 0 );
 700                 if( $limit < 0 ) {
 701                         $limit = 0;
 702                 }
 703                 if( ( $limit == 0 ) && ( $optionname != '' ) ) {
 704                         $limit = (int)$wgUser->getOption( $optionname );
 705                 }
 706                 if( $limit <= 0 ) {
 707                         $limit = $deflimit;
 708                 }
 709                 if( $limit > 5000 ) {
 710                         $limit = 5000; # We have *some* limits...
 711                 }
 712
 713                 $offset = $this->getInt( 'offset', 0 );
 714                 if( $offset < 0 ) {
 715                         $offset = 0;
 716                 }
 717
 718                 return array( $limit, $offset );
 719         }
 720
 721         /**
 722          * Return the path to the temporary file where PHP has stored the upload.
 723          *
 724          * @param $key String:
 725          * @return string or NULL if no such file.
 726          */
 727         public function getFileTempname( $key ) {
 728                 $file = new WebRequestUpload( $this, $key );
 729                 return $file->getTempName();
 730         }
 731
 732         /**
 733          * Return the size of the upload, or 0.
 734          *
 735          * @deprecated since 1.17
 736          * @param $key String:
 737          * @return integer
 738          */
 739         public function getFileSize( $key ) {
 740                 $file = new WebRequestUpload( $this, $key );
 741                 return $file->getSize();
 742         }
 743
 744         /**
 745          * Return the upload error or 0
 746          *
 747          * @param $key String:
 748          * @return integer
 749          */
 750         public function getUploadError( $key ) {
 751                 $file = new WebRequestUpload( $this, $key );
 752                 return $file->getError();
 753         }
 754
 755         /**
 756          * Return the original filename of the uploaded file, as reported by
 757          * the submitting user agent. HTML-style character entities are
 758          * interpreted and normalized to Unicode normalization form C, in part
 759          * to deal with weird input from Safari with non-ASCII filenames.
 760          *
 761          * Other than this the name is not verified for being a safe filename.
 762          *
 763          * @param $key String:
 764          * @return string or NULL if no such file.
 765          */
 766         public function getFileName( $key ) {
 767                 $file = new WebRequestUpload( $this, $key );
 768                 return $file->getName();
 769         }
 770
 771         /**
 772          * Return a WebRequestUpload object corresponding to the key
 773          *
 774          * @param $key string
 775          * @return WebRequestUpload
 776          */
 777         public function getUpload( $key ) {
 778                 return new WebRequestUpload( $this, $key );
 779         }
 780
 781         /**
 782          * Return a handle to WebResponse style object, for setting cookies,
 783          * headers and other stuff, for Request being worked on.
 784          *
 785          * @return WebResponse
 786          */
 787         public function response() {
 788                 /* Lazy initialization of response object for this request */
 789                 if ( !is_object( $this->response ) ) {
 790                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 791                         $this->response = new $class();
 792                 }
 793                 return $this->response;
 794         }
 795
 796         /**
 797          * Initialise the header list
 798          */
 799         private function initHeaders() {
 800                 if ( count( $this->headers ) ) {
 801                         return;
 802                 }
 803
 804                 if ( function_exists( 'apache_request_headers' ) ) {
 805                         foreach ( apache_request_headers() as $tempName => $tempValue ) {
 806                                 $this->headers[ strtoupper( $tempName ) ] = $tempValue;
 807                         }
 808                 } else {
 809                         foreach ( $_SERVER as $name => $value ) {
 810                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 811                                         $name = str_replace( '_', '-',  substr( $name, 5 ) );
 812                                         $this->headers[$name] = $value;
 813                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 814                                         $this->headers['CONTENT-LENGTH'] = $value;
 815                                 }
 816                         }
 817                 }
 818         }
 819
 820         /**
 821          * Get an array containing all request headers
 822          *
 823          * @return Array mapping header name to its value
 824          */
 825         public function getAllHeaders() {
 826                 $this->initHeaders();
 827                 return $this->headers;
 828         }
 829
 830         /**
 831          * Get a request header, or false if it isn't set
 832          * @param $name String: case-insensitive header name
 833          *
 834          * @return string|false
 835          */
 836         public function getHeader( $name ) {
 837                 $this->initHeaders();
 838                 $name = strtoupper( $name );
 839                 if ( isset( $this->headers[$name] ) ) {
 840                         return $this->headers[$name];
 841                 } else {
 842                         return false;
 843                 }
 844         }
 845
 846         /**
 847          * Get data from $_SESSION
 848          *
 849          * @param $key String: name of key in $_SESSION
 850          * @return Mixed
 851          */
 852         public function getSessionData( $key ) {
 853                 if( !isset( $_SESSION[$key] ) ) {
 854                         return null;
 855                 }
 856                 return $_SESSION[$key];
 857         }
 858
 859         /**
 860          * Set session data
 861          *
 862          * @param $key String: name of key in $_SESSION
 863          * @param $data Mixed
 864          */
 865         public function setSessionData( $key, $data ) {
 866                 $_SESSION[$key] = $data;
 867         }
 868
 869         /**
 870          * Check if Internet Explorer will detect an incorrect cache extension in
 871          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 872          * message or redirect to a safer URL. Returns true if the URL is OK, and
 873          * false if an error message has been shown and the request should be aborted.
 874          *
 875          * @param $extWhitelist array
 876          * @return bool
 877          */
 878         public function checkUrlExtension( $extWhitelist = array() ) {
 879                 global $wgScriptExtension;
 880                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 881                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 882                         if ( !$this->wasPosted() ) {
 883                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 884                                         $this->getFullRequestURL(), $extWhitelist );
 885                                 if ( $newUrl !== false ) {
 886                                         $this->doSecurityRedirect( $newUrl );
 887                                         return false;
 888                                 }
 889                         }
 890                         throw new HttpError( 403,
 891                                 'Invalid file extension found in the path info or query string.' );
 892                 }
 893                 return true;
 894         }
 895
 896         /**
 897          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 898          * IE 6. Returns true if it was successful, false otherwise.
 899          *
 900          * @param $url string
 901          * @return bool
 902          */
 903         protected function doSecurityRedirect( $url ) {
 904                 header( 'Location: ' . $url );
 905                 header( 'Content-Type: text/html' );
 906                 $encUrl = htmlspecialchars( $url );
 907                 echo <<<HTML
 908 <html>
 909 <head>
 910 <title>Security redirect</title>
 911 </head>
 912 <body>
 913 <h1>Security redirect</h1>
 914 <p>
 915 We can't serve non-HTML content from the URL you have requested, because
 916 Internet Explorer would interpret it as an incorrect and potentially dangerous
 917 content type.</p>
 918 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the URL you have requested, except that
 919 "&amp;*" is appended. This prevents Internet Explorer from seeing a bogus file
 920 extension.
 921 </p>
 922 </body>
 923 </html>
 924 HTML;
 925                 echo "\n";
 926                 return true;
 927         }
 928
 929         /**
 930          * Returns true if the PATH_INFO ends with an extension other than a script
 931          * extension. This could confuse IE for scripts that send arbitrary data which
 932          * is not HTML but may be detected as such.
 933          *
 934          * Various past attempts to use the URL to make this check have generally
 935          * run up against the fact that CGI does not provide a standard method to
 936          * determine the URL. PATH_INFO may be mangled (e.g. if cgi.fix_pathinfo=0),
 937          * but only by prefixing it with the script name and maybe some other stuff,
 938          * the extension is not mangled. So this should be a reasonably portable
 939          * way to perform this security check.
 940          *
 941          * Also checks for anything that looks like a file extension at the end of
 942          * QUERY_STRING, since IE 6 and earlier will use this to get the file type
 943          * if there was no dot before the question mark (bug 28235).
 944          *
 945          * @deprecated Use checkUrlExtension().
 946          *
 947          * @param $extWhitelist array
 948          *
 949          * @return bool
 950          */
 951         public function isPathInfoBad( $extWhitelist = array() ) {
 952                 global $wgScriptExtension;
 953                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 954                 return IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist );
 955         }
 956
 957         /**
 958          * Parse the Accept-Language header sent by the client into an array
 959          * @return array array( languageCode => q-value ) sorted by q-value in descending order
 960          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 961          * This is aligned with rfc2616 section 14.4
 962          */
 963         public function getAcceptLang() {
 964                 // Modified version of code found at http://www.thefutureoftheweb.com/blog/use-accept-language-header
 965                 $acceptLang = $this->getHeader( 'Accept-Language' );
 966                 if ( !$acceptLang ) {
 967                         return array();
 968                 }
 969
 970                 // Return the language codes in lower case
 971                 $acceptLang = strtolower( $acceptLang );
 972
 973                 // Break up string into pieces (languages and q factors)
 974                 $lang_parse = null;
 975                 preg_match_all( '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
 976                         $acceptLang, $lang_parse );
 977
 978                 if ( !count( $lang_parse[1] ) ) {
 979                         return array();
 980                 }
 981
 982                 // Create a list like "en" => 0.8
 983                 $langs = array_combine( $lang_parse[1], $lang_parse[4] );
 984                 // Set default q factor to 1
 985                 foreach ( $langs as $lang => $val ) {
 986                         if ( $val === '' ) {
 987                                 $langs[$lang] = 1;
 988                         } elseif ( $val == 0 ) {
 989                                 unset($langs[$lang]);
 990                         }
 991                 }
 992
 993                 // Sort list
 994                 arsort( $langs, SORT_NUMERIC );
 995                 return $langs;
 996         }
 997
 998         /**
 999          * Fetch the raw IP from the request
1000          *
1001          * @return String
1002          */
1003         protected function getRawIP() {
1004                 if ( isset( $_SERVER['REMOTE_ADDR'] ) ) {
1005                         return IP::canonicalize( $_SERVER['REMOTE_ADDR'] );
1006                 } else {
1007                         return null;
1008                 }
1009         }
1010
1011         /**
1012          * Work out the IP address based on various globals
1013          * For trusted proxies, use the XFF client IP (first of the chain)
1014          * @return string
1015          */
1016         public function getIP() {
1017                 global $wgUsePrivateIPs;
1018
1019                 # Return cached result
1020                 if ( $this->ip !== null ) {
1021                         return $this->ip;
1022                 }
1023
1024                 # collect the originating ips
1025                 $ip = $this->getRawIP();
1026
1027                 # Append XFF
1028                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1029                 if ( $forwardedFor !== false ) {
1030                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1031                         $ipchain = array_reverse( $ipchain );
1032                         if ( $ip ) {
1033                                 array_unshift( $ipchain, $ip );
1034                         }
1035
1036                         # Step through XFF list and find the last address in the list which is a trusted server
1037                         # Set $ip to the IP address given by that trusted server, unless the address is not sensible (e.g. private)
1038                         foreach ( $ipchain as $i => $curIP ) {
1039                                 $curIP = IP::canonicalize( $curIP );
1040                                 if ( wfIsTrustedProxy( $curIP ) ) {
1041                                         if ( isset( $ipchain[$i + 1] ) ) {
1042                                                 if ( $wgUsePrivateIPs || IP::isPublic( $ipchain[$i + 1 ] ) ) {
1043                                                         $ip = $ipchain[$i + 1];
1044                                                 }
1045                                         }
1046                                 } else {
1047                                         break;
1048                                 }
1049                         }
1050                 }
1051
1052                 # Allow extensions to improve our guess
1053                 wfRunHooks( 'GetIP', array( &$ip ) );
1054
1055                 if ( !$ip ) {
1056                         throw new MWException( "Unable to determine IP" );
1057                 }
1058
1059                 wfDebug( "IP: $ip\n" );
1060                 $this->ip = $ip;
1061                 return $ip;
1062         }
1063 }
1064
1065 /**
1066  * Object to access the $_FILES array
1067  */
1068 class WebRequestUpload {
1069         protected $request;
1070         protected $doesExist;
1071         protected $fileInfo;
1072
1073         /**
1074          * Constructor. Should only be called by WebRequest
1075          *
1076          * @param $request WebRequest The associated request
1077          * @param $key string Key in $_FILES array (name of form field)
1078          */
1079         public function __construct( $request, $key ) {
1080                 $this->request = $request;
1081                 $this->doesExist = isset( $_FILES[$key] );
1082                 if ( $this->doesExist ) {
1083                         $this->fileInfo = $_FILES[$key];
1084                 }
1085         }
1086
1087         /**
1088          * Return whether a file with this name was uploaded.
1089          *
1090          * @return bool
1091          */
1092         public function exists() {
1093                 return $this->doesExist;
1094         }
1095
1096         /**
1097          * Return the original filename of the uploaded file
1098          *
1099          * @return mixed Filename or null if non-existent
1100          */
1101         public function getName() {
1102                 if ( !$this->exists() ) {
1103                         return null;
1104                 }
1105
1106                 global $wgContLang;
1107                 $name = $this->fileInfo['name'];
1108
1109                 # Safari sends filenames in HTML-encoded Unicode form D...
1110                 # Horrid and evil! Let's try to make some kind of sense of it.
1111                 $name = Sanitizer::decodeCharReferences( $name );
1112                 $name = $wgContLang->normalize( $name );
1113                 wfDebug( __METHOD__ . ": {$this->fileInfo['name']} normalized to '$name'\n" );
1114                 return $name;
1115         }
1116
1117         /**
1118          * Return the file size of the uploaded file
1119          *
1120          * @return int File size or zero if non-existent
1121          */
1122         public function getSize() {
1123                 if ( !$this->exists() ) {
1124                         return 0;
1125                 }
1126
1127                 return $this->fileInfo['size'];
1128         }
1129
1130         /**
1131          * Return the path to the temporary file
1132          *
1133          * @return mixed Path or null if non-existent
1134          */
1135         public function getTempName() {
1136                 if ( !$this->exists() ) {
1137                         return null;
1138                 }
1139
1140                 return $this->fileInfo['tmp_name'];
1141         }
1142
1143         /**
1144          * Return the upload error. See link for explanation
1145          * http://www.php.net/manual/en/features.file-upload.errors.php
1146          *
1147          * @return int One of the UPLOAD_ constants, 0 if non-existent
1148          */
1149         public function getError() {
1150                 if ( !$this->exists() ) {
1151                         return 0; # UPLOAD_ERR_OK
1152                 }
1153
1154                 return $this->fileInfo['error'];
1155         }
1156
1157         /**
1158          * Returns whether this upload failed because of overflow of a maximum set
1159          * in php.ini
1160          *
1161          * @return bool
1162          */
1163         public function isIniSizeOverflow() {
1164                 if ( $this->getError() == UPLOAD_ERR_INI_SIZE ) {
1165                         # PHP indicated that upload_max_filesize is exceeded
1166                         return true;
1167                 }
1168
1169                 $contentLength = $this->request->getHeader( 'CONTENT_LENGTH' );
1170                 if ( $contentLength > wfShorthandToInteger( ini_get( 'post_max_size' ) ) ) {
1171                         # post_max_size is exceeded
1172                         return true;
1173                 }
1174
1175                 return false;
1176         }
1177 }
1178
1179 /**
1180  * WebRequest clone which takes values from a provided array.
1181  *
1182  * @ingroup HTTP
1183  */
1184 class FauxRequest extends WebRequest {
1185         private $wasPosted = false;
1186         private $session = array();
1187
1188         /**
1189          * @param $data Array of *non*-urlencoded key => value pairs, the
1190          *   fake GET/POST values
1191          * @param $wasPosted Bool: whether to treat the data as POST
1192          * @param $session Mixed: session array or null
1193          */
1194         public function __construct( $data, $wasPosted = false, $session = null ) {
1195                 if( is_array( $data ) ) {
1196                         $this->data = $data;
1197                 } else {
1198                         throw new MWException( "FauxRequest() got bogus data" );
1199                 }
1200                 $this->wasPosted = $wasPosted;
1201                 if( $session )
1202                         $this->session = $session;
1203         }
1204
1205         /**
1206          * @param $method string
1207          * @throws MWException
1208          */
1209         private function notImplemented( $method ) {
1210                 throw new MWException( "{$method}() not implemented" );
1211         }
1212
1213         /**
1214          * @param $name string
1215          * @param $default string
1216          * @return string
1217          */
1218         public function getText( $name, $default = '' ) {
1219                 # Override; don't recode since we're using internal data
1220                 return (string)$this->getVal( $name, $default );
1221         }
1222
1223         /**
1224          * @return Array
1225          */
1226         public function getValues() {
1227                 return $this->data;
1228         }
1229
1230         /**
1231          * @return array
1232          */
1233         public function getQueryValues() {
1234                 if ( $this->wasPosted ) {
1235                         return array();
1236                 } else {
1237                         return $this->data;
1238                 }
1239         }
1240
1241         /**
1242          * @return bool
1243          */
1244         public function wasPosted() {
1245                 return $this->wasPosted;
1246         }
1247
1248         public function checkSessionCookie() {
1249                 return false;
1250         }
1251
1252         public function getRequestURL() {
1253                 $this->notImplemented( __METHOD__ );
1254         }
1255
1256         /**
1257          * @param $name
1258          * @return bool|string
1259          */
1260         public function getHeader( $name ) {
1261                 return isset( $this->headers[$name] ) ? $this->headers[$name] : false;
1262         }
1263
1264         /**
1265          * @param $name string
1266          * @param $val string
1267          */
1268         public function setHeader( $name, $val ) {
1269                 $this->headers[$name] = $val;
1270         }
1271
1272         /**
1273          * @param $key
1274          * @return mixed
1275          */
1276         public function getSessionData( $key ) {
1277                 if( isset( $this->session[$key] ) )
1278                         return $this->session[$key];
1279         }
1280
1281         /**
1282          * @param $key
1283          * @param $data
1284          */
1285         public function setSessionData( $key, $data ) {
1286                 $this->session[$key] = $data;
1287         }
1288
1289         /**
1290          * @return array|Mixed|null
1291          */
1292         public function getSessionArray() {
1293                 return $this->session;
1294         }
1295
1296         /**
1297          * @param array $extWhitelist
1298          * @return bool
1299          */
1300         public function isPathInfoBad( $extWhitelist = array() ) {
1301                 return false;
1302         }
1303
1304         /**
1305          * @param array $extWhitelist
1306          * @return bool
1307          */
1308         public function checkUrlExtension( $extWhitelist = array() ) {
1309                 return true;
1310         }
1311
1312         /**
1313          * @return string
1314          */
1315         protected function getRawIP() {
1316                 return '127.0.0.1';
1317         }
1318 }
1319
1320 /**
1321  * Similar to FauxRequest, but only fakes URL parameters and method
1322  * (POST or GET) and use the base request for the remaining stuff
1323  * (cookies, session and headers).
1324  *
1325  * @ingroup HTTP
1326  */
1327 class DerivativeRequest extends FauxRequest {
1328         private $base;
1329
1330         public function __construct( WebRequest $base, $data, $wasPosted = false ) {
1331                 $this->base = $base;
1332                 parent::__construct( $data, $wasPosted );
1333         }
1334
1335         public function getCookie( $key, $prefix = null, $default = null ) {
1336                 return $this->base->getCookie( $key, $prefix, $default );
1337         }
1338
1339         public function checkSessionCookie() {
1340                 return $this->base->checkSessionCookie();
1341         }
1342
1343         public function getHeader( $name ) {
1344                 return $this->base->getHeader( $name );
1345         }
1346
1347         public function getAllHeaders() {
1348                 return $this->base->getAllHeaders();
1349         }
1350
1351         public function getSessionData( $key ) {
1352                 return $this->base->getSessionData( $key );
1353         }
1354
1355         public function setSessionData( $key, $data ) {
1356                 return $this->base->setSessionData( $key, $data );
1357         }
1358
1359         public function getAcceptLang() {
1360                 return $this->base->getAcceptLang();
1361         }
1362
1363         public function getIP() {
1364                 return $this->base->getIP();
1365         }
1366 }