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