includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasssty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * The WebRequest class encapsulates getting at data passed in the
  28  * URL or via a POSTed form, handling remove of "magic quotes" slashes,
  29  * stripping illegal input characters and normalizing Unicode sequences.
  30  *
  31  * Usually this is used via a global singleton, $wgRequest. You should
  32  * not create a second WebRequest object; make a FauxRequest object if
  33  * you want to pass arbitrary data to some function in place of the web
  34  * input.
  35  *
  36  * @ingroup HTTP
  37  */
  38 class WebRequest {
  39         protected $data, $headers = array();
  40
  41         /**
  42          * Lazy-init response object
  43          * @var WebResponse
  44          */
  45         private $response;
  46
  47         /**
  48          * Cached client IP address
  49          * @var String
  50          */
  51         private $ip;
  52
  53         public function __construct() {
  54                 /// @todo FIXME: This preemptive de-quoting can interfere with other web libraries
  55                 ///        and increases our memory footprint. It would be cleaner to do on
  56                 ///        demand; but currently we have no wrapper for $_SERVER etc.
  57                 $this->checkMagicQuotes();
  58
  59                 // POST overrides GET data
  60                 // We don't use $_REQUEST here to avoid interference from cookies...
  61                 $this->data = $_POST + $_GET;
  62         }
  63
  64         /**
  65          * Extract relevant query arguments from the http request uri's path
  66          * to be merged with the normal php provided query arguments.
  67          * Tries to use the REQUEST_URI data if available and parses it
  68          * according to the wiki's configuration looking for any known pattern.
  69          *
  70          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
  71          * provided by the server if any and use that to set a 'title' parameter.
  72          *
  73          * @param $want string: If this is not 'all', then the function
  74          * will return an empty array if it determines that the URL is
  75          * inside a rewrite path.
  76          *
  77          * @return Array: Any query arguments found in path matches.
  78          */
  79         static public function getPathInfo( $want = 'all' ) {
  80                 global $wgUsePathInfo;
  81                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
  82                 // And also by Apache 2.x, double slashes are converted to single slashes.
  83                 // So we will use REQUEST_URI if possible.
  84                 $matches = array();
  85                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
  86                         // Slurp out the path portion to examine...
  87                         $url = $_SERVER['REQUEST_URI'];
  88                         if ( !preg_match( '!^https?://!', $url ) ) {
  89                                 $url = 'http://unused' . $url;
  90                         }
  91                         $a = parse_url( $url );
  92                         if( $a ) {
  93                                 $path = isset( $a['path'] ) ? $a['path'] : '';
  94
  95                                 global $wgScript;
  96                                 if( $path == $wgScript && $want !== 'all' ) {
  97                                         // Script inside a rewrite path?
  98                                         // Abort to keep from breaking...
  99                                         return $matches;
 100                                 }
 101
 102                                 $router = new PathRouter;
 103
 104                                 // Raw PATH_INFO style
 105                                 $router->add( "$wgScript/$1" );
 106
 107                                 if( isset( $_SERVER['SCRIPT_NAME'] )
 108                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] ) )
 109                                 {
 110                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 111                                         # But we do have some other .php files such as img_auth.php
 112                                         # Don't let root article paths clober the parsing for them
 113                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 114                                 }
 115
 116                                 global $wgArticlePath;
 117                                 if( $wgArticlePath ) {
 118                                         $router->add( $wgArticlePath );
 119                                 }
 120
 121                                 global $wgActionPaths;
 122                                 if( $wgActionPaths ) {
 123                                         $router->add( $wgActionPaths, array( 'action' => '$key' ) );
 124                                 }
 125
 126                                 global $wgVariantArticlePath, $wgContLang;
 127                                 if( $wgVariantArticlePath ) {
 128                                         $router->add( $wgVariantArticlePath,
 129                                                 array( 'variant' => '$2'),
 130                                                 array( '$2' => $wgContLang->getVariants() )
 131                                         );
 132                                 }
 133
 134                                 wfRunHooks( 'WebRequestPathInfoRouter', array( $router ) );
 135
 136                                 $matches = $router->parse( $path );
 137                         }
 138                 } elseif ( $wgUsePathInfo ) {
 139                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 140                                 // Mangled PATH_INFO
 141                                 // http://bugs.php.net/bug.php?id=31892
 142                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 143                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 144
 145                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && ($_SERVER['PATH_INFO'] != '') ) {
 146                                 // Regular old PATH_INFO yay
 147                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 148                         }
 149                 }
 150
 151                 return $matches;
 152         }
 153
 154         /**
 155          * Work out an appropriate URL prefix containing scheme and host, based on
 156          * information detected from $_SERVER
 157          *
 158          * @return string
 159          */
 160         public static function detectServer() {
 161                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 162
 163                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 164                 $host = 'localhost';
 165                 $port = $stdPort;
 166                 foreach ( $varNames as $varName ) {
 167                         if ( !isset( $_SERVER[$varName] ) ) {
 168                                 continue;
 169                         }
 170                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 171                         if ( !$parts ) {
 172                                 // Invalid, do not use
 173                                 continue;
 174                         }
 175                         $host = $parts[0];
 176                         if ( $parts[1] === false ) {
 177                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 178                                         $port = $_SERVER['SERVER_PORT'];
 179                                 } // else leave it as $stdPort
 180                         } else {
 181                                 $port = $parts[1];
 182                         }
 183                         break;
 184                 }
 185
 186                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 187         }
 188
 189         /**
 190          * @return array
 191          */
 192         public static function detectProtocolAndStdPort() {
 193                 return ( isset( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] == 'on' ) ? array( 'https', 443 ) : array( 'http', 80 );
 194         }
 195
 196         /**
 197          * @return string
 198          */
 199         public static function detectProtocol() {
 200                 list( $proto, $stdPort ) = self::detectProtocolAndStdPort();
 201                 return $proto;
 202         }
 203
 204         /**
 205          * Check for title, action, and/or variant data in the URL
 206          * and interpolate it into the GET variables.
 207          * This should only be run after $wgContLang is available,
 208          * as we may need the list of language variants to determine
 209          * available variant URLs.
 210          */
 211         public function interpolateTitle() {
 212                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 213                 if ( defined( 'MW_API' ) ) {
 214                         return;
 215                 }
 216
 217                 $matches = self::getPathInfo( 'title' );
 218                 foreach( $matches as $key => $val) {
 219                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 220                 }
 221         }
 222
 223         /**
 224          * URL rewriting function; tries to extract page title and,
 225          * optionally, one other fixed parameter value from a URL path.
 226          *
 227          * @param $path string: the URL path given from the client
 228          * @param $bases array: one or more URLs, optionally with $1 at the end
 229          * @param $key string: if provided, the matching key in $bases will be
 230          *             passed on as the value of this URL parameter
 231          * @return array of URL variables to interpolate; empty if no match
 232          */
 233         static function extractTitle( $path, $bases, $key = false ) {
 234                 foreach( (array)$bases as $keyValue => $base ) {
 235                         // Find the part after $wgArticlePath
 236                         $base = str_replace( '$1', '', $base );
 237                         $baseLen = strlen( $base );
 238                         if( substr( $path, 0, $baseLen ) == $base ) {
 239                                 $raw = substr( $path, $baseLen );
 240                                 if( $raw !== '' ) {
 241                                         $matches = array( 'title' => rawurldecode( $raw ) );
 242                                         if( $key ) {
 243                                                 $matches[$key] = $keyValue;
 244                                         }
 245                                         return $matches;
 246                                 }
 247                         }
 248                 }
 249                 return array();
 250         }
 251
 252         /**
 253          * Recursively strips slashes from the given array;
 254          * used for undoing the evil that is magic_quotes_gpc.
 255          *
 256          * @param $arr array: will be modified
 257          * @param $topLevel bool Specifies if the array passed is from the top
 258          * level of the source. In PHP5 magic_quotes only escapes the first level
 259          * of keys that belong to an array.
 260          * @return array the original array
 261          * @see http://www.php.net/manual/en/function.get-magic-quotes-gpc.php#49612
 262          */
 263         private function &fix_magic_quotes( &$arr, $topLevel = true ) {
 264                 $clean = array();
 265                 foreach( $arr as $key => $val ) {
 266                         if( is_array( $val ) ) {
 267                                 $cleanKey = $topLevel ? stripslashes( $key ) : $key;
 268                                 $clean[$cleanKey] = $this->fix_magic_quotes( $arr[$key], false );
 269                         } else {
 270                                 $cleanKey = stripslashes( $key );
 271                                 $clean[$cleanKey] = stripslashes( $val );
 272                         }
 273                 }
 274                 $arr = $clean;
 275                 return $arr;
 276         }
 277
 278         /**
 279          * If magic_quotes_gpc option is on, run the global arrays
 280          * through fix_magic_quotes to strip out the stupid slashes.
 281          * WARNING: This should only be done once! Running a second
 282          * time could damage the values.
 283          */
 284         private function checkMagicQuotes() {
 285                 $mustFixQuotes = function_exists( 'get_magic_quotes_gpc' )
 286                         && get_magic_quotes_gpc();
 287                 if( $mustFixQuotes ) {
 288                         $this->fix_magic_quotes( $_COOKIE );
 289                         $this->fix_magic_quotes( $_ENV );
 290                         $this->fix_magic_quotes( $_GET );
 291                         $this->fix_magic_quotes( $_POST );
 292                         $this->fix_magic_quotes( $_REQUEST );
 293                         $this->fix_magic_quotes( $_SERVER );
 294                 }
 295         }
 296
 297         /**
 298          * Recursively normalizes UTF-8 strings in the given array.
 299          *
 300          * @param $data string|array
 301          * @return array|string cleaned-up version of the given
 302          * @private
 303          */
 304         function normalizeUnicode( $data ) {
 305                 if( is_array( $data ) ) {
 306                         foreach( $data as $key => $val ) {
 307                                 $data[$key] = $this->normalizeUnicode( $val );
 308                         }
 309                 } else {
 310                         global $wgContLang;
 311                         $data = isset( $wgContLang ) ? $wgContLang->normalize( $data ) : UtfNormal::cleanUp( $data );
 312                 }
 313                 return $data;
 314         }
 315
 316         /**
 317          * Fetch a value from the given array or return $default if it's not set.
 318          *
 319          * @param $arr Array
 320          * @param $name String
 321          * @param $default Mixed
 322          * @return mixed
 323          */
 324         private function getGPCVal( $arr, $name, $default ) {
 325                 # PHP is so nice to not touch input data, except sometimes:
 326                 # http://us2.php.net/variables.external#language.variables.external.dot-in-names
 327                 # Work around PHP *feature* to avoid *bugs* elsewhere.
 328                 $name = strtr( $name, '.', '_' );
 329                 if( isset( $arr[$name] ) ) {
 330                         global $wgContLang;
 331                         $data = $arr[$name];
 332                         if( isset( $_GET[$name] ) && !is_array( $data ) ) {
 333                                 # Check for alternate/legacy character encoding.
 334                                 if( isset( $wgContLang ) ) {
 335                                         $data = $wgContLang->checkTitleEncoding( $data );
 336                                 }
 337                         }
 338                         $data = $this->normalizeUnicode( $data );
 339                         return $data;
 340                 } else {
 341                         taint( $default );
 342                         return $default;
 343                 }
 344         }
 345
 346         /**
 347          * Fetch a scalar from the input or return $default if it's not set.
 348          * Returns a string. Arrays are discarded. Useful for
 349          * non-freeform text inputs (e.g. predefined internal text keys
 350          * selected by a drop-down menu). For freeform input, see getText().
 351          *
 352          * @param $name String
 353          * @param $default String: optional default (or NULL)
 354          * @return String
 355          */
 356         public function getVal( $name, $default = null ) {
 357                 $val = $this->getGPCVal( $this->data, $name, $default );
 358                 if( is_array( $val ) ) {
 359                         $val = $default;
 360                 }
 361                 if( is_null( $val ) ) {
 362                         return $val;
 363                 } else {
 364                         return (string)$val;
 365                 }
 366         }
 367
 368         /**
 369          * Set an arbitrary value into our get/post data.
 370          *
 371          * @param $key String: key name to use
 372          * @param $value Mixed: value to set
 373          * @return Mixed: old value if one was present, null otherwise
 374          */
 375         public function setVal( $key, $value ) {
 376                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 377                 $this->data[$key] = $value;
 378                 return $ret;
 379         }
 380
 381
 382         /**
 383          * Unset an arbitrary value from our get/post data.
 384          *
 385          * @param $key String: key name to use
 386          * @return Mixed: old value if one was present, null otherwise
 387          */
 388         public function unsetVal( $key ) {
 389                 if ( !isset( $this->data[$key] ) ) {
 390                         $ret = null;
 391                 } else {
 392                         $ret = $this->data[$key];
 393                         unset( $this->data[$key] );
 394                 }
 395                 return $ret;
 396         }
 397
 398         /**
 399          * Fetch an array from the input or return $default if it's not set.
 400          * If source was scalar, will return an array with a single element.
 401          * If no source and no default, returns NULL.
 402          *
 403          * @param $name String
 404          * @param $default Array: optional default (or NULL)
 405          * @return Array
 406          */
 407         public function getArray( $name, $default = null ) {
 408                 $val = $this->getGPCVal( $this->data, $name, $default );
 409                 if( is_null( $val ) ) {
 410                         return null;
 411                 } else {
 412                         return (array)$val;
 413                 }
 414         }
 415
 416         /**
 417          * Fetch an array of integers, or return $default if it's not set.
 418          * If source was scalar, will return an array with a single element.
 419          * If no source and no default, returns NULL.
 420          * If an array is returned, contents are guaranteed to be integers.
 421          *
 422          * @param $name String
 423          * @param $default Array: option default (or NULL)
 424          * @return Array of ints
 425          */
 426         public function getIntArray( $name, $default = null ) {
 427                 $val = $this->getArray( $name, $default );
 428                 if( is_array( $val ) ) {
 429                         $val = array_map( 'intval', $val );
 430                 }
 431                 return $val;
 432         }
 433
 434         /**
 435          * Fetch an integer value from the input or return $default if not set.
 436          * Guaranteed to return an integer; non-numeric input will typically
 437          * return 0.
 438          *
 439          * @param $name String
 440          * @param $default Integer
 441          * @return Integer
 442          */
 443         public function getInt( $name, $default = 0 ) {
 444                 return intval( $this->getVal( $name, $default ) );
 445         }
 446
 447         /**
 448          * Fetch an integer value from the input or return null if empty.
 449          * Guaranteed to return an integer or null; non-numeric input will
 450          * typically return null.
 451          *
 452          * @param $name String
 453          * @return Integer
 454          */
 455         public function getIntOrNull( $name ) {
 456                 $val = $this->getVal( $name );
 457                 return is_numeric( $val )
 458                         ? intval( $val )
 459                         : null;
 460         }
 461
 462         /**
 463          * Fetch a boolean value from the input or return $default if not set.
 464          * Guaranteed to return true or false, with normal PHP semantics for
 465          * boolean interpretation of strings.
 466          *
 467          * @param $name String
 468          * @param $default Boolean
 469          * @return Boolean
 470          */
 471         public function getBool( $name, $default = false ) {
 472                 return (bool)$this->getVal( $name, $default );
 473         }
 474
 475         /**
 476          * Fetch a boolean value from the input or return $default if not set.
 477          * Unlike getBool, the string "false" will result in boolean false, which is
 478          * useful when interpreting information sent from JavaScript.
 479          *
 480          * @param $name String
 481          * @param $default Boolean
 482          * @return Boolean
 483          */
 484         public function getFuzzyBool( $name, $default = false ) {
 485                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 486         }
 487
 488         /**
 489          * Return true if the named value is set in the input, whatever that
 490          * value is (even "0"). Return false if the named value is not set.
 491          * Example use is checking for the presence of check boxes in forms.
 492          *
 493          * @param $name String
 494          * @return Boolean
 495          */
 496         public function getCheck( $name ) {
 497                 # Checkboxes and buttons are only present when clicked
 498                 # Presence connotes truth, abscense false
 499                 $val = $this->getVal( $name, null );
 500                 return isset( $val );
 501         }
 502
 503         /**
 504          * Fetch a text string from the given array or return $default if it's not
 505          * set. Carriage returns are stripped from the text, and with some language
 506          * modules there is an input transliteration applied. This should generally
 507          * be used for form <textarea> and <input> fields. Used for user-supplied
 508          * freeform text input (for which input transformations may be required - e.g.
 509          * Esperanto x-coding).
 510          *
 511          * @param $name String
 512          * @param $default String: optional
 513          * @return String
 514          */
 515         public function getText( $name, $default = '' ) {
 516                 global $wgContLang;
 517                 $val = $this->getVal( $name, $default );
 518                 return str_replace( "\r\n", "\n",
 519                         $wgContLang->recodeInput( $val ) );
 520         }
 521
 522         /**
 523          * Extracts the given named values into an array.
 524          * If no arguments are given, returns all input values.
 525          * No transformation is performed on the values.
 526          *
 527          * @return array
 528          */
 529         public function getValues() {
 530                 $names = func_get_args();
 531                 if ( count( $names ) == 0 ) {
 532                         $names = array_keys( $this->data );
 533                 }
 534
 535                 $retVal = array();
 536                 foreach ( $names as $name ) {
 537                         $value = $this->getGPCVal( $this->data, $name, null );
 538                         if ( !is_null( $value ) ) {
 539                                 $retVal[$name] = $value;
 540                         }
 541                 }
 542                 return $retVal;
 543         }
 544
 545         /**
 546          * Returns the names of all input values excluding those in $exclude.
 547          *
 548          * @param $exclude Array
 549          * @return array
 550          */
 551         public function getValueNames( $exclude = array() ) {
 552                 return array_diff( array_keys( $this->getValues() ), $exclude );
 553         }
 554
 555         /**
 556          * Get the values passed in the query string.
 557          * No transformation is performed on the values.
 558          *
 559          * @return Array
 560          */
 561          public function getQueryValues() {
 562                 return $_GET;
 563          }
 564
 565         /**
 566          * Returns true if the present request was reached by a POST operation,
 567          * false otherwise (GET, HEAD, or command-line).
 568          *
 569          * Note that values retrieved by the object may come from the
 570          * GET URL etc even on a POST request.
 571          *
 572          * @return Boolean
 573          */
 574         public function wasPosted() {
 575                 return isset( $_SERVER['REQUEST_METHOD'] ) && $_SERVER['REQUEST_METHOD'] == 'POST';
 576         }
 577
 578         /**
 579          * Returns true if there is a session cookie set.
 580          * This does not necessarily mean that the user is logged in!
 581          *
 582          * If you want to check for an open session, use session_id()
 583          * instead; that will also tell you if the session was opened
 584          * during the current request (in which case the cookie will
 585          * be sent back to the client at the end of the script run).
 586          *
 587          * @return Boolean
 588          */
 589         public function checkSessionCookie() {
 590                 return isset( $_COOKIE[ session_name() ] );
 591         }
 592
 593         /**
 594          * Get a cookie from the $_COOKIE jar
 595          *
 596          * @param $key String: the name of the cookie
 597          * @param $prefix String: a prefix to use for the cookie name, if not $wgCookiePrefix
 598          * @param $default Mixed: what to return if the value isn't found
 599          * @return Mixed: cookie value or $default if the cookie not set
 600          */
 601         public function getCookie( $key, $prefix = null, $default = null ) {
 602                 if( $prefix === null ) {
 603                         global $wgCookiePrefix;
 604                         $prefix = $wgCookiePrefix;
 605                 }
 606                 return $this->getGPCVal( $_COOKIE, $prefix . $key , $default );
 607         }
 608
 609         /**
 610          * Return the path and query string portion of the request URI.
 611          * This will be suitable for use as a relative link in HTML output.
 612          *
 613          * @return String
 614          */
 615         public function getRequestURL() {
 616                 if( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 617                         $base = $_SERVER['REQUEST_URI'];
 618                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] ) && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] ) ) {
 619                         // Probably IIS; doesn't set REQUEST_URI
 620                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 621                 } elseif( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 622                         $base = $_SERVER['SCRIPT_NAME'];
 623                         if( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 624                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 625                         }
 626                 } else {
 627                         // This shouldn't happen!
 628                         throw new MWException( "Web server doesn't provide either " .
 629                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 630                                 "of your web server configuration to http://bugzilla.wikimedia.org/" );
 631                 }
 632                 // User-agents should not send a fragment with the URI, but
 633                 // if they do, and the web server passes it on to us, we
 634                 // need to strip it or we get false-positive redirect loops
 635                 // or weird output URLs
 636                 $hash = strpos( $base, '#' );
 637                 if( $hash !== false ) {
 638                         $base = substr( $base, 0, $hash );
 639                 }
 640                 if( $base[0] == '/' ) {
 641                         return $base;
 642                 } else {
 643                         // We may get paths with a host prepended; strip it.
 644                         return preg_replace( '!^[^:]+://[^/]+/!', '/', $base );
 645                 }
 646         }
 647
 648         /**
 649          * Return the request URI with the canonical service and hostname, path,
 650          * and query string. This will be suitable for use as an absolute link
 651          * in HTML or other output.
 652          *
 653          * If $wgServer is protocol-relative, this will return a fully
 654          * qualified URL with the protocol that was used for this request.
 655          *
 656          * @return String
 657          */
 658         public function getFullRequestURL() {
 659                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 660         }
 661
 662         /**
 663          * Take an arbitrary query and rewrite the present URL to include it
 664          * @param $query String: query string fragment; do not include initial '?'
 665          *
 666          * @return String
 667          */
 668         public function appendQuery( $query ) {
 669                 return $this->appendQueryArray( wfCgiToArray( $query ) );
 670         }
 671
 672         /**
 673          * HTML-safe version of appendQuery().
 674          * @deprecated: Deprecated in 1.20, warnings in 1.21, remove in 1.22.
 675          *
 676          * @param $query String: query string fragment; do not include initial '?'
 677          * @return String
 678          */
 679         public function escapeAppendQuery( $query ) {
 680                 return htmlspecialchars( $this->appendQuery( $query ) );
 681         }
 682
 683         /**
 684          * @param $key
 685          * @param $value
 686          * @param $onlyquery bool
 687          * @return String
 688          */
 689         public function appendQueryValue( $key, $value, $onlyquery = false ) {
 690                 return $this->appendQueryArray( array( $key => $value ), $onlyquery );
 691         }
 692
 693         /**
 694          * Appends or replaces value of query variables.
 695          *
 696          * @param $array Array of values to replace/add to query
 697          * @param $onlyquery Bool: whether to only return the query string and not
 698          *                   the complete URL
 699          * @return String
 700          */
 701         public function appendQueryArray( $array, $onlyquery = false ) {
 702                 global $wgTitle;
 703                 $newquery = $this->getQueryValues();
 704                 unset( $newquery['title'] );
 705                 $newquery = array_merge( $newquery, $array );
 706                 $query = wfArrayToCGI( $newquery );
 707                 return $onlyquery ? $query : $wgTitle->getLocalURL( $query );
 708         }
 709
 710         /**
 711          * Check for limit and offset parameters on the input, and return sensible
 712          * defaults if not given. The limit must be positive and is capped at 5000.
 713          * Offset must be positive but is not capped.
 714          *
 715          * @param $deflimit Integer: limit to use if no input and the user hasn't set the option.
 716          * @param $optionname String: to specify an option other than rclimit to pull from.
 717          * @return array first element is limit, second is offset
 718          */
 719         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 720                 global $wgUser;
 721
 722                 $limit = $this->getInt( 'limit', 0 );
 723                 if( $limit < 0 ) {
 724                         $limit = 0;
 725                 }
 726                 if( ( $limit == 0 ) && ( $optionname != '' ) ) {
 727                         $limit = (int)$wgUser->getOption( $optionname );
 728                 }
 729                 if( $limit <= 0 ) {
 730                         $limit = $deflimit;
 731                 }
 732                 if( $limit > 5000 ) {
 733                         $limit = 5000; # We have *some* limits...
 734                 }
 735
 736                 $offset = $this->getInt( 'offset', 0 );
 737                 if( $offset < 0 ) {
 738                         $offset = 0;
 739                 }
 740
 741                 return array( $limit, $offset );
 742         }
 743
 744         /**
 745          * Return the path to the temporary file where PHP has stored the upload.
 746          *
 747          * @param $key String:
 748          * @return string or NULL if no such file.
 749          */
 750         public function getFileTempname( $key ) {
 751                 $file = new WebRequestUpload( $this, $key );
 752                 return $file->getTempName();
 753         }
 754
 755         /**
 756          * Return the size of the upload, or 0.
 757          *
 758          * @deprecated since 1.17
 759          * @param $key String:
 760          * @return integer
 761          */
 762         public function getFileSize( $key ) {
 763                 wfDeprecated( __METHOD__, '1.17' );
 764                 $file = new WebRequestUpload( $this, $key );
 765                 return $file->getSize();
 766         }
 767
 768         /**
 769          * Return the upload error or 0
 770          *
 771          * @param $key String:
 772          * @return integer
 773          */
 774         public function getUploadError( $key ) {
 775                 $file = new WebRequestUpload( $this, $key );
 776                 return $file->getError();
 777         }
 778
 779         /**
 780          * Return the original filename of the uploaded file, as reported by
 781          * the submitting user agent. HTML-style character entities are
 782          * interpreted and normalized to Unicode normalization form C, in part
 783          * to deal with weird input from Safari with non-ASCII filenames.
 784          *
 785          * Other than this the name is not verified for being a safe filename.
 786          *
 787          * @param $key String:
 788          * @return string or NULL if no such file.
 789          */
 790         public function getFileName( $key ) {
 791                 $file = new WebRequestUpload( $this, $key );
 792                 return $file->getName();
 793         }
 794
 795         /**
 796          * Return a WebRequestUpload object corresponding to the key
 797          *
 798          * @param $key string
 799          * @return WebRequestUpload
 800          */
 801         public function getUpload( $key ) {
 802                 return new WebRequestUpload( $this, $key );
 803         }
 804
 805         /**
 806          * Return a handle to WebResponse style object, for setting cookies,
 807          * headers and other stuff, for Request being worked on.
 808          *
 809          * @return WebResponse
 810          */
 811         public function response() {
 812                 /* Lazy initialization of response object for this request */
 813                 if ( !is_object( $this->response ) ) {
 814                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 815                         $this->response = new $class();
 816                 }
 817                 return $this->response;
 818         }
 819
 820         /**
 821          * Initialise the header list
 822          */
 823         private function initHeaders() {
 824                 if ( count( $this->headers ) ) {
 825                         return;
 826                 }
 827
 828                 if ( function_exists( 'apache_request_headers' ) ) {
 829                         foreach ( apache_request_headers() as $tempName => $tempValue ) {
 830                                 $this->headers[ strtoupper( $tempName ) ] = $tempValue;
 831                         }
 832                 } else {
 833                         foreach ( $_SERVER as $name => $value ) {
 834                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 835                                         $name = str_replace( '_', '-',  substr( $name, 5 ) );
 836                                         $this->headers[$name] = $value;
 837                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 838                                         $this->headers['CONTENT-LENGTH'] = $value;
 839                                 }
 840                         }
 841                 }
 842         }
 843
 844         /**
 845          * Get an array containing all request headers
 846          *
 847          * @return Array mapping header name to its value
 848          */
 849         public function getAllHeaders() {
 850                 $this->initHeaders();
 851                 return $this->headers;
 852         }
 853
 854         /**
 855          * Get a request header, or false if it isn't set
 856          * @param $name String: case-insensitive header name
 857          *
 858          * @return string|bool False on failure
 859          */
 860         public function getHeader( $name ) {
 861                 $this->initHeaders();
 862                 $name = strtoupper( $name );
 863                 if ( isset( $this->headers[$name] ) ) {
 864                         return $this->headers[$name];
 865                 } else {
 866                         return false;
 867                 }
 868         }
 869
 870         /**
 871          * Get data from $_SESSION
 872          *
 873          * @param $key String: name of key in $_SESSION
 874          * @return Mixed
 875          */
 876         public function getSessionData( $key ) {
 877                 if( !isset( $_SESSION[$key] ) ) {
 878                         return null;
 879                 }
 880                 return $_SESSION[$key];
 881         }
 882
 883         /**
 884          * Set session data
 885          *
 886          * @param $key String: name of key in $_SESSION
 887          * @param $data Mixed
 888          */
 889         public function setSessionData( $key, $data ) {
 890                 $_SESSION[$key] = $data;
 891         }
 892
 893         /**
 894          * Check if Internet Explorer will detect an incorrect cache extension in
 895          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 896          * message or redirect to a safer URL. Returns true if the URL is OK, and
 897          * false if an error message has been shown and the request should be aborted.
 898          *
 899          * @param $extWhitelist array
 900          * @return bool
 901          */
 902         public function checkUrlExtension( $extWhitelist = array() ) {
 903                 global $wgScriptExtension;
 904                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 905                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 906                         if ( !$this->wasPosted() ) {
 907                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 908                                         $this->getFullRequestURL(), $extWhitelist );
 909                                 if ( $newUrl !== false ) {
 910                                         $this->doSecurityRedirect( $newUrl );
 911                                         return false;
 912                                 }
 913                         }
 914                         throw new HttpError( 403,
 915                                 'Invalid file extension found in the path info or query string.' );
 916                 }
 917                 return true;
 918         }
 919
 920         /**
 921          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 922          * IE 6. Returns true if it was successful, false otherwise.
 923          *
 924          * @param $url string
 925          * @return bool
 926          */
 927         protected function doSecurityRedirect( $url ) {
 928                 header( 'Location: ' . $url );
 929                 header( 'Content-Type: text/html' );
 930                 $encUrl = htmlspecialchars( $url );
 931                 echo <<<HTML
 932 <html>
 933 <head>
 934 <title>Security redirect</title>
 935 </head>
 936 <body>
 937 <h1>Security redirect</h1>
 938 <p>
 939 We can't serve non-HTML content from the URL you have requested, because
 940 Internet Explorer would interpret it as an incorrect and potentially dangerous
 941 content type.</p>
 942 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the URL you have requested, except that
 943 "&amp;*" is appended. This prevents Internet Explorer from seeing a bogus file
 944 extension.
 945 </p>
 946 </body>
 947 </html>
 948 HTML;
 949                 echo "\n";
 950                 return true;
 951         }
 952
 953         /**
 954          * Returns true if the PATH_INFO ends with an extension other than a script
 955          * extension. This could confuse IE for scripts that send arbitrary data which
 956          * is not HTML but may be detected as such.
 957          *
 958          * Various past attempts to use the URL to make this check have generally
 959          * run up against the fact that CGI does not provide a standard method to
 960          * determine the URL. PATH_INFO may be mangled (e.g. if cgi.fix_pathinfo=0),
 961          * but only by prefixing it with the script name and maybe some other stuff,
 962          * the extension is not mangled. So this should be a reasonably portable
 963          * way to perform this security check.
 964          *
 965          * Also checks for anything that looks like a file extension at the end of
 966          * QUERY_STRING, since IE 6 and earlier will use this to get the file type
 967          * if there was no dot before the question mark (bug 28235).
 968          *
 969          * @deprecated Use checkUrlExtension().
 970          *
 971          * @param $extWhitelist array
 972          *
 973          * @return bool
 974          */
 975         public function isPathInfoBad( $extWhitelist = array() ) {
 976                 wfDeprecated( __METHOD__, '1.17' );
 977                 global $wgScriptExtension;
 978                 $extWhitelist[] = ltrim( $wgScriptExtension, '.' );
 979                 return IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist );
 980         }
 981
 982         /**
 983          * Parse the Accept-Language header sent by the client into an array
 984          * @return array array( languageCode => q-value ) sorted by q-value in descending order
 985          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 986          * This is aligned with rfc2616 section 14.4
 987          */
 988         public function getAcceptLang() {
 989                 // Modified version of code found at http://www.thefutureoftheweb.com/blog/use-accept-language-header
 990                 $acceptLang = $this->getHeader( 'Accept-Language' );
 991                 if ( !$acceptLang ) {
 992                         return array();
 993                 }
 994
 995                 // Return the language codes in lower case
 996                 $acceptLang = strtolower( $acceptLang );
 997
 998                 // Break up string into pieces (languages and q factors)
 999                 $lang_parse = null;
1000                 preg_match_all( '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1001                         $acceptLang, $lang_parse );
1002
1003                 if ( !count( $lang_parse[1] ) ) {
1004                         return array();
1005                 }
1006
1007                 // Create a list like "en" => 0.8
1008                 $langs = array_combine( $lang_parse[1], $lang_parse[4] );
1009                 // Set default q factor to 1
1010                 foreach ( $langs as $lang => $val ) {
1011                         if ( $val === '' ) {
1012                                 $langs[$lang] = 1;
1013                         } elseif ( $val == 0 ) {
1014                                 unset($langs[$lang]);
1015                         }
1016                 }
1017
1018                 // Sort list
1019                 arsort( $langs, SORT_NUMERIC );
1020                 return $langs;
1021         }
1022
1023         /**
1024          * Fetch the raw IP from the request
1025          *
1026          * @since 1.19
1027          *
1028          * @return String
1029          */
1030         protected function getRawIP() {
1031                 if ( isset( $_SERVER['REMOTE_ADDR'] ) ) {
1032                         return IP::canonicalize( $_SERVER['REMOTE_ADDR'] );
1033                 } else {
1034                         return null;
1035                 }
1036         }
1037
1038         /**
1039          * Work out the IP address based on various globals
1040          * For trusted proxies, use the XFF client IP (first of the chain)
1041          *
1042          * @since 1.19
1043          *
1044          * @return string
1045          */
1046         public function getIP() {
1047                 global $wgUsePrivateIPs;
1048
1049                 # Return cached result
1050                 if ( $this->ip !== null ) {
1051                         return $this->ip;
1052                 }
1053
1054                 # collect the originating ips
1055                 $ip = $this->getRawIP();
1056
1057                 # Append XFF
1058                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1059                 if ( $forwardedFor !== false ) {
1060                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1061                         $ipchain = array_reverse( $ipchain );
1062                         if ( $ip ) {
1063                                 array_unshift( $ipchain, $ip );
1064                         }
1065
1066                         # Step through XFF list and find the last address in the list which is a trusted server
1067                         # Set $ip to the IP address given by that trusted server, unless the address is not sensible (e.g. private)
1068                         foreach ( $ipchain as $i => $curIP ) {
1069                                 $curIP = IP::canonicalize( $curIP );
1070                                 if ( wfIsTrustedProxy( $curIP ) ) {
1071                                         if ( isset( $ipchain[$i + 1] ) ) {
1072                                                 if ( $wgUsePrivateIPs || IP::isPublic( $ipchain[$i + 1 ] ) ) {
1073                                                         $ip = $ipchain[$i + 1];
1074                                                 }
1075                                         }
1076                                 } else {
1077                                         break;
1078                                 }
1079                         }
1080                 }
1081
1082                 # Allow extensions to improve our guess
1083                 wfRunHooks( 'GetIP', array( &$ip ) );
1084
1085                 if ( !$ip ) {
1086                         throw new MWException( "Unable to determine IP" );
1087                 }
1088
1089                 wfDebug( "IP: $ip\n" );
1090                 $this->ip = $ip;
1091                 return $ip;
1092         }
1093 }
1094
1095 /**
1096  * Object to access the $_FILES array
1097  */
1098 class WebRequestUpload {
1099         protected $request;
1100         protected $doesExist;
1101         protected $fileInfo;
1102
1103         /**
1104          * Constructor. Should only be called by WebRequest
1105          *
1106          * @param $request WebRequest The associated request
1107          * @param $key string Key in $_FILES array (name of form field)
1108          */
1109         public function __construct( $request, $key ) {
1110                 $this->request = $request;
1111                 $this->doesExist = isset( $_FILES[$key] );
1112                 if ( $this->doesExist ) {
1113                         $this->fileInfo = $_FILES[$key];
1114                 }
1115         }
1116
1117         /**
1118          * Return whether a file with this name was uploaded.
1119          *
1120          * @return bool
1121          */
1122         public function exists() {
1123                 return $this->doesExist;
1124         }
1125
1126         /**
1127          * Return the original filename of the uploaded file
1128          *
1129          * @return mixed Filename or null if non-existent
1130          */
1131         public function getName() {
1132                 if ( !$this->exists() ) {
1133                         return null;
1134                 }
1135
1136                 global $wgContLang;
1137                 $name = $this->fileInfo['name'];
1138
1139                 # Safari sends filenames in HTML-encoded Unicode form D...
1140                 # Horrid and evil! Let's try to make some kind of sense of it.
1141                 $name = Sanitizer::decodeCharReferences( $name );
1142                 $name = $wgContLang->normalize( $name );
1143                 wfDebug( __METHOD__ . ": {$this->fileInfo['name']} normalized to '$name'\n" );
1144                 return $name;
1145         }
1146
1147         /**
1148          * Return the file size of the uploaded file
1149          *
1150          * @return int File size or zero if non-existent
1151          */
1152         public function getSize() {
1153                 if ( !$this->exists() ) {
1154                         return 0;
1155                 }
1156
1157                 return $this->fileInfo['size'];
1158         }
1159
1160         /**
1161          * Return the path to the temporary file
1162          *
1163          * @return mixed Path or null if non-existent
1164          */
1165         public function getTempName() {
1166                 if ( !$this->exists() ) {
1167                         return null;
1168                 }
1169
1170                 return $this->fileInfo['tmp_name'];
1171         }
1172
1173         /**
1174          * Return the upload error. See link for explanation
1175          * http://www.php.net/manual/en/features.file-upload.errors.php
1176          *
1177          * @return int One of the UPLOAD_ constants, 0 if non-existent
1178          */
1179         public function getError() {
1180                 if ( !$this->exists() ) {
1181                         return 0; # UPLOAD_ERR_OK
1182                 }
1183
1184                 return $this->fileInfo['error'];
1185         }
1186
1187         /**
1188          * Returns whether this upload failed because of overflow of a maximum set
1189          * in php.ini
1190          *
1191          * @return bool
1192          */
1193         public function isIniSizeOverflow() {
1194                 if ( $this->getError() == UPLOAD_ERR_INI_SIZE ) {
1195                         # PHP indicated that upload_max_filesize is exceeded
1196                         return true;
1197                 }
1198
1199                 $contentLength = $this->request->getHeader( 'CONTENT_LENGTH' );
1200                 if ( $contentLength > wfShorthandToInteger( ini_get( 'post_max_size' ) ) ) {
1201                         # post_max_size is exceeded
1202                         return true;
1203                 }
1204
1205                 return false;
1206         }
1207 }
1208
1209 /**
1210  * WebRequest clone which takes values from a provided array.
1211  *
1212  * @ingroup HTTP
1213  */
1214 class FauxRequest extends WebRequest {
1215         private $wasPosted = false;
1216         private $session = array();
1217
1218         /**
1219          * @param $data Array of *non*-urlencoded key => value pairs, the
1220          *   fake GET/POST values
1221          * @param $wasPosted Bool: whether to treat the data as POST
1222          * @param $session Mixed: session array or null
1223          */
1224         public function __construct( $data = array(), $wasPosted = false, $session = null ) {
1225                 if( is_array( $data ) ) {
1226                         $this->data = $data;
1227                 } else {
1228                         throw new MWException( "FauxRequest() got bogus data" );
1229                 }
1230                 $this->wasPosted = $wasPosted;
1231                 if( $session )
1232                         $this->session = $session;
1233         }
1234
1235         /**
1236          * @param $method string
1237          * @throws MWException
1238          */
1239         private function notImplemented( $method ) {
1240                 throw new MWException( "{$method}() not implemented" );
1241         }
1242
1243         /**
1244          * @param $name string
1245          * @param $default string
1246          * @return string
1247          */
1248         public function getText( $name, $default = '' ) {
1249                 # Override; don't recode since we're using internal data
1250                 return (string)$this->getVal( $name, $default );
1251         }
1252
1253         /**
1254          * @return Array
1255          */
1256         public function getValues() {
1257                 return $this->data;
1258         }
1259
1260         /**
1261          * @return array
1262          */
1263         public function getQueryValues() {
1264                 if ( $this->wasPosted ) {
1265                         return array();
1266                 } else {
1267                         return $this->data;
1268                 }
1269         }
1270
1271         /**
1272          * @return bool
1273          */
1274         public function wasPosted() {
1275                 return $this->wasPosted;
1276         }
1277
1278         public function checkSessionCookie() {
1279                 return false;
1280         }
1281
1282         public function getRequestURL() {
1283                 $this->notImplemented( __METHOD__ );
1284         }
1285
1286         /**
1287          * @param $name
1288          * @return bool|string
1289          */
1290         public function getHeader( $name ) {
1291                 return isset( $this->headers[$name] ) ? $this->headers[$name] : false;
1292         }
1293
1294         /**
1295          * @param $name string
1296          * @param $val string
1297          */
1298         public function setHeader( $name, $val ) {
1299                 $this->headers[$name] = $val;
1300         }
1301
1302         /**
1303          * @param $key
1304          * @return mixed
1305          */
1306         public function getSessionData( $key ) {
1307                 if( isset( $this->session[$key] ) )
1308                         return $this->session[$key];
1309         }
1310
1311         /**
1312          * @param $key
1313          * @param $data
1314          */
1315         public function setSessionData( $key, $data ) {
1316                 $this->session[$key] = $data;
1317         }
1318
1319         /**
1320          * @return array|Mixed|null
1321          */
1322         public function getSessionArray() {
1323                 return $this->session;
1324         }
1325
1326         /**
1327          * @param array $extWhitelist
1328          * @return bool
1329          */
1330         public function isPathInfoBad( $extWhitelist = array() ) {
1331                 return false;
1332         }
1333
1334         /**
1335          * @param array $extWhitelist
1336          * @return bool
1337          */
1338         public function checkUrlExtension( $extWhitelist = array() ) {
1339                 return true;
1340         }
1341
1342         /**
1343          * @return string
1344          */
1345         protected function getRawIP() {
1346                 return '127.0.0.1';
1347         }
1348 }
1349
1350 /**
1351  * Similar to FauxRequest, but only fakes URL parameters and method
1352  * (POST or GET) and use the base request for the remaining stuff
1353  * (cookies, session and headers).
1354  *
1355  * @ingroup HTTP
1356  * @since 1.19
1357  */
1358 class DerivativeRequest extends FauxRequest {
1359         private $base;
1360
1361         public function __construct( WebRequest $base, $data, $wasPosted = false ) {
1362                 $this->base = $base;
1363                 parent::__construct( $data, $wasPosted );
1364         }
1365
1366         public function getCookie( $key, $prefix = null, $default = null ) {
1367                 return $this->base->getCookie( $key, $prefix, $default );
1368         }
1369
1370         public function checkSessionCookie() {
1371                 return $this->base->checkSessionCookie();
1372         }
1373
1374         public function getHeader( $name ) {
1375                 return $this->base->getHeader( $name );
1376         }
1377
1378         public function getAllHeaders() {
1379                 return $this->base->getAllHeaders();
1380         }
1381
1382         public function getSessionData( $key ) {
1383                 return $this->base->getSessionData( $key );
1384         }
1385
1386         public function setSessionData( $key, $data ) {
1387                 $this->base->setSessionData( $key, $data );
1388         }
1389
1390         public function getAcceptLang() {
1391                 return $this->base->getAcceptLang();
1392         }
1393
1394         public function getIP() {
1395                 return $this->base->getIP();
1396         }
1397 }