includes/WebRequest.php

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