includes/WebRequest.php

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