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