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          * Get the values passed in the query string.
 490          * No transformation is performed on the values.
 491          *
 492          * @return Array
 493          */
 494          public function getQueryValues() {
 495                 return $_GET;
 496          }
 497
 498         /**
 499          * Returns true if the present request was reached by a POST operation,
 500          * false otherwise (GET, HEAD, or command-line).
 501          *
 502          * Note that values retrieved by the object may come from the
 503          * GET URL etc even on a POST request.
 504          *
 505          * @return Boolean
 506          */
 507         public function wasPosted() {
 508                 return $_SERVER['REQUEST_METHOD'] == 'POST';
 509         }
 510
 511         /**
 512          * Returns true if there is a session cookie set.
 513          * This does not necessarily mean that the user is logged in!
 514          *
 515          * If you want to check for an open session, use session_id()
 516          * instead; that will also tell you if the session was opened
 517          * during the current request (in which case the cookie will
 518          * be sent back to the client at the end of the script run).
 519          *
 520          * @return Boolean
 521          */
 522         public function checkSessionCookie() {
 523                 return isset( $_COOKIE[ session_name() ] );
 524         }
 525
 526         /**
 527          * Get a cookie from the $_COOKIE jar
 528          *
 529          * @param $key String: the name of the cookie
 530          * @param $prefix String: a prefix to use for the cookie name, if not $wgCookiePrefix
 531          * @param $default Mixed: what to return if the value isn't found
 532          * @return Mixed: cookie value or $default if the cookie not set
 533          */
 534         public function getCookie( $key, $prefix = null, $default = null ) {
 535                 if( $prefix === null ) {
 536                         global $wgCookiePrefix;
 537                         $prefix = $wgCookiePrefix;
 538                 }
 539                 return $this->getGPCVal( $_COOKIE, $prefix . $key , $default );
 540         }
 541
 542         /**
 543          * Return the path and query string portion of the request URI.
 544          * This will be suitable for use as a relative link in HTML output.
 545          *
 546          * @return String
 547          */
 548         public function getRequestURL() {
 549                 if( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 550                         $base = $_SERVER['REQUEST_URI'];
 551                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] ) && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] ) ) {
 552                         // Probably IIS; doesn't set REQUEST_URI
 553                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 554                 } elseif( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 555                         $base = $_SERVER['SCRIPT_NAME'];
 556                         if( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 557                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 558                         }
 559                 } else {
 560                         // This shouldn't happen!
 561                         throw new MWException( "Web server doesn't provide either " .
 562                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 563                                 "of your web server configuration to http://bugzilla.wikimedia.org/" );
 564                 }
 565                 // User-agents should not send a fragment with the URI, but
 566                 // if they do, and the web server passes it on to us, we
 567                 // need to strip it or we get false-positive redirect loops
 568                 // or weird output URLs
 569                 $hash = strpos( $base, '#' );
 570                 if( $hash !== false ) {
 571                         $base = substr( $base, 0, $hash );
 572                 }
 573                 if( $base[0] == '/' ) {
 574                         return $base;
 575                 } else {
 576                         // We may get paths with a host prepended; strip it.
 577                         return preg_replace( '!^[^:]+://[^/]+/!', '/', $base );
 578                 }
 579         }
 580
 581         /**
 582          * Return the request URI with the canonical service and hostname, path,
 583          * and query string. This will be suitable for use as an absolute link
 584          * in HTML or other output.
 585          *
 586          * @return String
 587          */
 588         public function getFullRequestURL() {
 589                 global $wgServer;
 590                 return $wgServer . $this->getRequestURL();
 591         }
 592
 593         /**
 594          * Take an arbitrary query and rewrite the present URL to include it
 595          * @param $query String: query string fragment; do not include initial '?'
 596          *
 597          * @return String
 598          */
 599         public function appendQuery( $query ) {
 600                 return $this->appendQueryArray( wfCgiToArray( $query ) );
 601         }
 602
 603         /**
 604          * HTML-safe version of appendQuery().
 605          *
 606          * @param $query String: query string fragment; do not include initial '?'
 607          * @return String
 608          */
 609         public function escapeAppendQuery( $query ) {
 610                 return htmlspecialchars( $this->appendQuery( $query ) );
 611         }
 612
 613         /**
 614          * @param $key
 615          * @param $value
 616          * @param $onlyquery bool
 617          * @return String
 618          */
 619         public function appendQueryValue( $key, $value, $onlyquery = false ) {
 620                 return $this->appendQueryArray( array( $key => $value ), $onlyquery );
 621         }
 622
 623         /**
 624          * Appends or replaces value of query variables.
 625          *
 626          * @param $array Array of values to replace/add to query
 627          * @param $onlyquery Bool: whether to only return the query string and not
 628          *                   the complete URL
 629          * @return String
 630          */
 631         public function appendQueryArray( $array, $onlyquery = false ) {
 632                 global $wgTitle;
 633                 $newquery = $this->getQueryValues();
 634                 unset( $newquery['title'] );
 635                 $newquery = array_merge( $newquery, $array );
 636                 $query = wfArrayToCGI( $newquery );
 637                 return $onlyquery ? $query : $wgTitle->getLocalURL( $query );
 638         }
 639
 640         /**
 641          * Check for limit and offset parameters on the input, and return sensible
 642          * defaults if not given. The limit must be positive and is capped at 5000.
 643          * Offset must be positive but is not capped.
 644          *
 645          * @param $deflimit Integer: limit to use if no input and the user hasn't set the option.
 646          * @param $optionname String: to specify an option other than rclimit to pull from.
 647          * @return array first element is limit, second is offset
 648          */
 649         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 650                 global $wgUser;
 651
 652                 $limit = $this->getInt( 'limit', 0 );
 653                 if( $limit < 0 ) {
 654                         $limit = 0;
 655                 }
 656                 if( ( $limit == 0 ) && ( $optionname != '' ) ) {
 657                         $limit = (int)$wgUser->getOption( $optionname );
 658                 }
 659                 if( $limit <= 0 ) {
 660                         $limit = $deflimit;
 661                 }
 662                 if( $limit > 5000 ) {
 663                         $limit = 5000; # We have *some* limits...
 664                 }
 665
 666                 $offset = $this->getInt( 'offset', 0 );
 667                 if( $offset < 0 ) {
 668                         $offset = 0;
 669                 }
 670
 671                 return array( $limit, $offset );
 672         }
 673
 674         /**
 675          * Return the path to the temporary file where PHP has stored the upload.
 676          *
 677          * @param $key String:
 678          * @return string or NULL if no such file.
 679          */
 680         public function getFileTempname( $key ) {
 681                 $file = new WebRequestUpload( $this, $key );
 682                 return $file->getTempName();
 683         }
 684
 685         /**
 686          * Return the size of the upload, or 0.
 687          *
 688          * @deprecated since 1.17
 689          * @param $key String:
 690          * @return integer
 691          */
 692         public function getFileSize( $key ) {
 693                 $file = new WebRequestUpload( $this, $key );
 694                 return $file->getSize();
 695         }
 696
 697         /**
 698          * Return the upload error or 0
 699          *
 700          * @param $key String:
 701          * @return integer
 702          */
 703         public function getUploadError( $key ) {
 704                 $file = new WebRequestUpload( $this, $key );
 705                 return $file->getError();
 706         }
 707
 708         /**
 709          * Return the original filename of the uploaded file, as reported by
 710          * the submitting user agent. HTML-style character entities are
 711          * interpreted and normalized to Unicode normalization form C, in part
 712          * to deal with weird input from Safari with non-ASCII filenames.
 713          *
 714          * Other than this the name is not verified for being a safe filename.
 715          *
 716          * @param $key String:
 717          * @return string or NULL if no such file.
 718          */
 719         public function getFileName( $key ) {
 720                 $file = new WebRequestUpload( $this, $key );
 721                 return $file->getName();
 722         }
 723
 724         /**
 725          * Return a WebRequestUpload object corresponding to the key
 726          *
 727          * @param $key string
 728          * @return WebRequestUpload
 729          */
 730         public function getUpload( $key ) {
 731                 return new WebRequestUpload( $this, $key );
 732         }
 733
 734         /**
 735          * Return a handle to WebResponse style object, for setting cookies,
 736          * headers and other stuff, for Request being worked on.
 737          *
 738          * @return WebResponse
 739          */
 740         public function response() {
 741                 /* Lazy initialization of response object for this request */
 742                 if ( !is_object( $this->response ) ) {
 743                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 744                         $this->response = new $class();
 745                 }
 746                 return $this->response;
 747         }
 748
 749         /**
 750          * Initialise the header list
 751          */
 752         private function initHeaders() {
 753                 if ( count( $this->headers ) ) {
 754                         return;
 755                 }
 756
 757                 if ( function_exists( 'apache_request_headers' ) ) {
 758                         foreach ( apache_request_headers() as $tempName => $tempValue ) {
 759                                 $this->headers[ strtoupper( $tempName ) ] = $tempValue;
 760                         }
 761                 } else {
 762                         foreach ( $_SERVER as $name => $value ) {
 763                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 764                                         $name = str_replace( '_', '-',  substr( $name, 5 ) );
 765                                         $this->headers[$name] = $value;
 766                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 767                                         $this->headers['CONTENT-LENGTH'] = $value;
 768                                 }
 769                         }
 770                 }
 771         }
 772
 773         /**
 774          * Get an array containing all request headers
 775          *
 776          * @return Array mapping header name to its value
 777          */
 778         public function getAllHeaders() {
 779                 $this->initHeaders();
 780                 return $this->headers;
 781         }
 782
 783         /**
 784          * Get a request header, or false if it isn't set
 785          * @param $name String: case-insensitive header name
 786          *
 787          * @return string|false
 788          */
 789         public function getHeader( $name ) {
 790                 $this->initHeaders();
 791                 $name = strtoupper( $name );
 792                 if ( isset( $this->headers[$name] ) ) {
 793                         return $this->headers[$name];
 794                 } else {
 795                         return false;
 796                 }
 797         }
 798
 799         /**
 800          * Get data from $_SESSION
 801          *
 802          * @param $key String: name of key in $_SESSION
 803          * @return Mixed
 804          */
 805         public function getSessionData( $key ) {
 806                 if( !isset( $_SESSION[$key] ) ) {
 807                         return null;
 808                 }
 809                 return $_SESSION[$key];
 810         }
 811
 812         /**
 813          * Set session data
 814          *
 815          * @param $key String: name of key in $_SESSION
 816          * @param $data Mixed
 817          */
 818         public function setSessionData( $key, $data ) {
 819                 $_SESSION[$key] = $data;
 820         }
 821
 822         /**
 823          * Check if Internet Explorer will detect an incorrect cache extension in
 824          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 825          * message or redirect to a safer URL. Returns true if the URL is OK, and
 826          * false if an error message has been shown and the request should be aborted.
 827          *
 828          * @param $extWhitelist array
 829          * @return bool
 830          */
 831         public function checkUrlExtension( $extWhitelist = array() ) {
 832                 global $wgScriptExtension;
 833                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 834                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 835                         if ( !$this->wasPosted() ) {
 836                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 837                                         $this->getFullRequestURL(), $extWhitelist );
 838                                 if ( $newUrl !== false ) {
 839                                         $this->doSecurityRedirect( $newUrl );
 840                                         return false;
 841                                 }
 842                         }
 843                         wfHttpError( 403, 'Forbidden',
 844                                 'Invalid file extension found in the path info or query string.' );
 845
 846                         return false;
 847                 }
 848                 return true;
 849         }
 850
 851         /**
 852          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 853          * IE 6. Returns true if it was successful, false otherwise.
 854          *
 855          * @param $url string
 856          * @return bool
 857          */
 858         protected function doSecurityRedirect( $url ) {
 859                 header( 'Location: ' . $url );
 860                 header( 'Content-Type: text/html' );
 861                 $encUrl = htmlspecialchars( $url );
 862                 echo <<<HTML
 863 <html>
 864 <head>
 865 <title>Security redirect</title>
 866 </head>
 867 <body>
 868 <h1>Security redirect</h1>
 869 <p>
 870 We can't serve non-HTML content from the URL you have requested, because
 871 Internet Explorer would interpret it as an incorrect and potentially dangerous
 872 content type.</p>
 873 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the URL you have requested, except that
 874 "&amp;*" is appended. This prevents Internet Explorer from seeing a bogus file
 875 extension.
 876 </p>
 877 </body>
 878 </html>
 879 HTML;
 880                 echo "\n";
 881                 return true;
 882         }
 883
 884         /**
 885          * Returns true if the PATH_INFO ends with an extension other than a script
 886          * extension. This could confuse IE for scripts that send arbitrary data which
 887          * is not HTML but may be detected as such.
 888          *
 889          * Various past attempts to use the URL to make this check have generally
 890          * run up against the fact that CGI does not provide a standard method to
 891          * determine the URL. PATH_INFO may be mangled (e.g. if cgi.fix_pathinfo=0),
 892          * but only by prefixing it with the script name and maybe some other stuff,
 893          * the extension is not mangled. So this should be a reasonably portable
 894          * way to perform this security check.
 895          *
 896          * Also checks for anything that looks like a file extension at the end of
 897          * QUERY_STRING, since IE 6 and earlier will use this to get the file type
 898          * if there was no dot before the question mark (bug 28235).
 899          *
 900          * @deprecated Use checkUrlExtension().
 901          */
 902         public function isPathInfoBad( $extWhitelist = array() ) {
 903                 global $wgScriptExtension;
 904                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 905                 return IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist );
 906         }
 907
 908         /**
 909          * Parse the Accept-Language header sent by the client into an array
 910          * @return array( languageCode => q-value ) sorted by q-value in descending order
 911          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 912          * This is aligned with rfc2616 section 14.4
 913          */
 914         public function getAcceptLang() {
 915                 // Modified version of code found at http://www.thefutureoftheweb.com/blog/use-accept-language-header
 916                 $acceptLang = $this->getHeader( 'Accept-Language' );
 917                 if ( !$acceptLang ) {
 918                         return array();
 919                 }
 920
 921                 // Return the language codes in lower case
 922                 $acceptLang = strtolower( $acceptLang );
 923
 924                 // Break up string into pieces (languages and q factors)
 925                 $lang_parse = null;
 926                 preg_match_all( '/([a-z]{1,8}(-[a-z]{1,8})?|\*)\s*(;\s*q\s*=\s*(1|0(\.[0-9]+)?)?)?/',
 927                         $acceptLang, $lang_parse );
 928
 929                 if ( !count( $lang_parse[1] ) ) {
 930                         return array();
 931                 }
 932
 933                 // Create a list like "en" => 0.8
 934                 $langs = array_combine( $lang_parse[1], $lang_parse[4] );
 935                 // Set default q factor to 1
 936                 foreach ( $langs as $lang => $val ) {
 937                         if ( $val === '' ) {
 938                                 $langs[$lang] = 1;
 939                         } elseif ( $val == 0 ) {
 940                                 unset($langs[$lang]);
 941                         }
 942                 }
 943
 944                 // Sort list
 945                 arsort( $langs, SORT_NUMERIC );
 946                 return $langs;
 947         }
 948 }
 949
 950 /**
 951  * Object to access the $_FILES array
 952  */
 953 class WebRequestUpload {
 954         protected $request;
 955         protected $doesExist;
 956         protected $fileInfo;
 957
 958         /**
 959          * Constructor. Should only be called by WebRequest
 960          *
 961          * @param $request WebRequest The associated request
 962          * @param $key string Key in $_FILES array (name of form field)
 963          */
 964         public function __construct( $request, $key ) {
 965                 $this->request = $request;
 966                 $this->doesExist = isset( $_FILES[$key] );
 967                 if ( $this->doesExist ) {
 968                         $this->fileInfo = $_FILES[$key];
 969                 }
 970         }
 971
 972         /**
 973          * Return whether a file with this name was uploaded.
 974          *
 975          * @return bool
 976          */
 977         public function exists() {
 978                 return $this->doesExist;
 979         }
 980
 981         /**
 982          * Return the original filename of the uploaded file
 983          *
 984          * @return mixed Filename or null if non-existent
 985          */
 986         public function getName() {
 987                 if ( !$this->exists() ) {
 988                         return null;
 989                 }
 990
 991                 global $wgContLang;
 992                 $name = $this->fileInfo['name'];
 993
 994                 # Safari sends filenames in HTML-encoded Unicode form D...
 995                 # Horrid and evil! Let's try to make some kind of sense of it.
 996                 $name = Sanitizer::decodeCharReferences( $name );
 997                 $name = $wgContLang->normalize( $name );
 998                 wfDebug( __METHOD__ . ": {$this->fileInfo['name']} normalized to '$name'\n" );
 999                 return $name;
1000         }
1001
1002         /**
1003          * Return the file size of the uploaded file
1004          *
1005          * @return int File size or zero if non-existent
1006          */
1007         public function getSize() {
1008                 if ( !$this->exists() ) {
1009                         return 0;
1010                 }
1011
1012                 return $this->fileInfo['size'];
1013         }
1014
1015         /**
1016          * Return the path to the temporary file
1017          *
1018          * @return mixed Path or null if non-existent
1019          */
1020         public function getTempName() {
1021                 if ( !$this->exists() ) {
1022                         return null;
1023                 }
1024
1025                 return $this->fileInfo['tmp_name'];
1026         }
1027
1028         /**
1029          * Return the upload error. See link for explanation
1030          * http://www.php.net/manual/en/features.file-upload.errors.php
1031          *
1032          * @return int One of the UPLOAD_ constants, 0 if non-existent
1033          */
1034         public function getError() {
1035                 if ( !$this->exists() ) {
1036                         return 0; # UPLOAD_ERR_OK
1037                 }
1038
1039                 return $this->fileInfo['error'];
1040         }
1041
1042         /**
1043          * Returns whether this upload failed because of overflow of a maximum set
1044          * in php.ini
1045          *
1046          * @return bool
1047          */
1048         public function isIniSizeOverflow() {
1049                 if ( $this->getError() == UPLOAD_ERR_INI_SIZE ) {
1050                         # PHP indicated that upload_max_filesize is exceeded
1051                         return true;
1052                 }
1053
1054                 $contentLength = $this->request->getHeader( 'CONTENT_LENGTH' );
1055                 if ( $contentLength > wfShorthandToInteger( ini_get( 'post_max_size' ) ) ) {
1056                         # post_max_size is exceeded
1057                         return true;
1058                 }
1059
1060                 return false;
1061         }
1062 }
1063
1064 /**
1065  * WebRequest clone which takes values from a provided array.
1066  *
1067  * @ingroup HTTP
1068  */
1069 class FauxRequest extends WebRequest {
1070         private $wasPosted = false;
1071         private $session = array();
1072
1073         /**
1074          * @param $data Array of *non*-urlencoded key => value pairs, the
1075          *   fake GET/POST values
1076          * @param $wasPosted Bool: whether to treat the data as POST
1077          * @param $session Mixed: session array or null
1078          */
1079         public function __construct( $data, $wasPosted = false, $session = null ) {
1080                 if( is_array( $data ) ) {
1081                         $this->data = $data;
1082                 } else {
1083                         throw new MWException( "FauxRequest() got bogus data" );
1084                 }
1085                 $this->wasPosted = $wasPosted;
1086                 if( $session )
1087                         $this->session = $session;
1088         }
1089
1090         private function notImplemented( $method ) {
1091                 throw new MWException( "{$method}() not implemented" );
1092         }
1093
1094         public function getText( $name, $default = '' ) {
1095                 # Override; don't recode since we're using internal data
1096                 return (string)$this->getVal( $name, $default );
1097         }
1098
1099         public function getValues() {
1100                 return $this->data;
1101         }
1102
1103         public function getQueryValues() {
1104                 if ( $this->wasPosted ) {
1105                         return array();
1106                 } else {
1107                         return $this->data;
1108                 }
1109         }
1110
1111         public function wasPosted() {
1112                 return $this->wasPosted;
1113         }
1114
1115         public function checkSessionCookie() {
1116                 return false;
1117         }
1118
1119         public function getRequestURL() {
1120                 $this->notImplemented( __METHOD__ );
1121         }
1122
1123         public function getHeader( $name ) {
1124                 return isset( $this->headers[$name] ) ? $this->headers[$name] : false;
1125         }
1126
1127         public function setHeader( $name, $val ) {
1128                 $this->headers[$name] = $val;
1129         }
1130
1131         public function getSessionData( $key ) {
1132                 if( isset( $this->session[$key] ) )
1133                         return $this->session[$key];
1134         }
1135
1136         public function setSessionData( $key, $data ) {
1137                 $this->session[$key] = $data;
1138         }
1139
1140         public function getSessionArray() {
1141                 return $this->session;
1142         }
1143
1144         public function isPathInfoBad( $extWhitelist = array() ) {
1145                 return false;
1146         }
1147
1148         public function checkUrlExtension( $extWhitelist = array() ) {
1149                 return true;
1150         }
1151 }