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