includes/WebRequest.php

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