includes/WebRequest.php

   1 <?php
   2 /**
   3  * Deal with importing all those nasty globals and things
   4  *
   5  * Copyright © 2003 Brion Vibber <brion@pobox.com>
   6  * https://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 stripping illegal input characters and
  29  * normalizing Unicode sequences.
  30  *
  31  * @ingroup HTTP
  32  */
  33 class WebRequest {
  34         protected $data, $headers = array();
  35
  36         /**
  37          * Flag to make WebRequest::getHeader return an array of values.
  38          * @since 1.26
  39          */
  40         const GETHEADER_LIST = 1;
  41
  42         /**
  43          * Lazy-init response object
  44          * @var WebResponse
  45          */
  46         private $response;
  47
  48         /**
  49          * Cached client IP address
  50          * @var string
  51          */
  52         private $ip;
  53
  54         /**
  55          * The timestamp of the start of the request, with microsecond precision.
  56          * @var float
  57          */
  58         protected $requestTime;
  59
  60         /**
  61          * Cached URL protocol
  62          * @var string
  63          */
  64         protected $protocol;
  65
  66         public function __construct() {
  67                 $this->requestTime = isset( $_SERVER['REQUEST_TIME_FLOAT'] )
  68                         ? $_SERVER['REQUEST_TIME_FLOAT'] : microtime( true );
  69
  70                 // POST overrides GET data
  71                 // We don't use $_REQUEST here to avoid interference from cookies...
  72                 $this->data = $_POST + $_GET;
  73         }
  74
  75         /**
  76          * Extract relevant query arguments from the http request uri's path
  77          * to be merged with the normal php provided query arguments.
  78          * Tries to use the REQUEST_URI data if available and parses it
  79          * according to the wiki's configuration looking for any known pattern.
  80          *
  81          * If the REQUEST_URI is not provided we'll fall back on the PATH_INFO
  82          * provided by the server if any and use that to set a 'title' parameter.
  83          *
  84          * @param string $want If this is not 'all', then the function
  85          * will return an empty array if it determines that the URL is
  86          * inside a rewrite path.
  87          *
  88          * @return array Any query arguments found in path matches.
  89          */
  90         public static function getPathInfo( $want = 'all' ) {
  91                 global $wgUsePathInfo;
  92                 // PATH_INFO is mangled due to http://bugs.php.net/bug.php?id=31892
  93                 // And also by Apache 2.x, double slashes are converted to single slashes.
  94                 // So we will use REQUEST_URI if possible.
  95                 $matches = array();
  96                 if ( !empty( $_SERVER['REQUEST_URI'] ) ) {
  97                         // Slurp out the path portion to examine...
  98                         $url = $_SERVER['REQUEST_URI'];
  99                         if ( !preg_match( '!^https?://!', $url ) ) {
 100                                 $url = 'http://unused' . $url;
 101                         }
 102                         MediaWiki\suppressWarnings();
 103                         $a = parse_url( $url );
 104                         MediaWiki\restoreWarnings();
 105                         if ( $a ) {
 106                                 $path = isset( $a['path'] ) ? $a['path'] : '';
 107
 108                                 global $wgScript;
 109                                 if ( $path == $wgScript && $want !== 'all' ) {
 110                                         // Script inside a rewrite path?
 111                                         // Abort to keep from breaking...
 112                                         return $matches;
 113                                 }
 114
 115                                 $router = new PathRouter;
 116
 117                                 // Raw PATH_INFO style
 118                                 $router->add( "$wgScript/$1" );
 119
 120                                 if ( isset( $_SERVER['SCRIPT_NAME'] )
 121                                         && preg_match( '/\.php5?/', $_SERVER['SCRIPT_NAME'] )
 122                                 ) {
 123                                         # Check for SCRIPT_NAME, we handle index.php explicitly
 124                                         # But we do have some other .php files such as img_auth.php
 125                                         # Don't let root article paths clober the parsing for them
 126                                         $router->add( $_SERVER['SCRIPT_NAME'] . "/$1" );
 127                                 }
 128
 129                                 global $wgArticlePath;
 130                                 if ( $wgArticlePath ) {
 131                                         $router->add( $wgArticlePath );
 132                                 }
 133
 134                                 global $wgActionPaths;
 135                                 if ( $wgActionPaths ) {
 136                                         $router->add( $wgActionPaths, array( 'action' => '$key' ) );
 137                                 }
 138
 139                                 global $wgVariantArticlePath, $wgContLang;
 140                                 if ( $wgVariantArticlePath ) {
 141                                         $router->add( $wgVariantArticlePath,
 142                                                 array( 'variant' => '$2' ),
 143                                                 array( '$2' => $wgContLang->getVariants() )
 144                                         );
 145                                 }
 146
 147                                 Hooks::run( 'WebRequestPathInfoRouter', array( $router ) );
 148
 149                                 $matches = $router->parse( $path );
 150                         }
 151                 } elseif ( $wgUsePathInfo ) {
 152                         if ( isset( $_SERVER['ORIG_PATH_INFO'] ) && $_SERVER['ORIG_PATH_INFO'] != '' ) {
 153                                 // Mangled PATH_INFO
 154                                 // http://bugs.php.net/bug.php?id=31892
 155                                 // Also reported when ini_get('cgi.fix_pathinfo')==false
 156                                 $matches['title'] = substr( $_SERVER['ORIG_PATH_INFO'], 1 );
 157
 158                         } elseif ( isset( $_SERVER['PATH_INFO'] ) && $_SERVER['PATH_INFO'] != '' ) {
 159                                 // Regular old PATH_INFO yay
 160                                 $matches['title'] = substr( $_SERVER['PATH_INFO'], 1 );
 161                         }
 162                 }
 163
 164                 return $matches;
 165         }
 166
 167         /**
 168          * Work out an appropriate URL prefix containing scheme and host, based on
 169          * information detected from $_SERVER
 170          *
 171          * @return string
 172          */
 173         public static function detectServer() {
 174                 global $wgAssumeProxiesUseDefaultProtocolPorts;
 175
 176                 $proto = self::detectProtocol();
 177                 $stdPort = $proto === 'https' ? 443 : 80;
 178
 179                 $varNames = array( 'HTTP_HOST', 'SERVER_NAME', 'HOSTNAME', 'SERVER_ADDR' );
 180                 $host = 'localhost';
 181                 $port = $stdPort;
 182                 foreach ( $varNames as $varName ) {
 183                         if ( !isset( $_SERVER[$varName] ) ) {
 184                                 continue;
 185                         }
 186
 187                         $parts = IP::splitHostAndPort( $_SERVER[$varName] );
 188                         if ( !$parts ) {
 189                                 // Invalid, do not use
 190                                 continue;
 191                         }
 192
 193                         $host = $parts[0];
 194                         if ( $wgAssumeProxiesUseDefaultProtocolPorts && isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) ) {
 195                                 // Bug 70021: Assume that upstream proxy is running on the default
 196                                 // port based on the protocol. We have no reliable way to determine
 197                                 // the actual port in use upstream.
 198                                 $port = $stdPort;
 199                         } elseif ( $parts[1] === false ) {
 200                                 if ( isset( $_SERVER['SERVER_PORT'] ) ) {
 201                                         $port = $_SERVER['SERVER_PORT'];
 202                                 } // else leave it as $stdPort
 203                         } else {
 204                                 $port = $parts[1];
 205                         }
 206                         break;
 207                 }
 208
 209                 return $proto . '://' . IP::combineHostAndPort( $host, $port, $stdPort );
 210         }
 211
 212         /**
 213          * Detect the protocol from $_SERVER.
 214          * This is for use prior to Setup.php, when no WebRequest object is available.
 215          * At other times, use the non-static function getProtocol().
 216          *
 217          * @return array
 218          */
 219         public static function detectProtocol() {
 220                 if ( ( !empty( $_SERVER['HTTPS'] ) && $_SERVER['HTTPS'] !== 'off' ) ||
 221                         ( isset( $_SERVER['HTTP_X_FORWARDED_PROTO'] ) &&
 222                         $_SERVER['HTTP_X_FORWARDED_PROTO'] === 'https' ) ) {
 223                         return 'https';
 224                 } else {
 225                         return 'http';
 226                 }
 227         }
 228
 229         /**
 230          * Get the number of seconds to have elapsed since request start,
 231          * in fractional seconds, with microsecond resolution.
 232          *
 233          * @return float
 234          * @since 1.25
 235          */
 236         public function getElapsedTime() {
 237                 return microtime( true ) - $this->requestTime;
 238         }
 239
 240         /**
 241          * Get the current URL protocol (http or https)
 242          * @return string
 243          */
 244         public function getProtocol() {
 245                 if ( $this->protocol === null ) {
 246                         $this->protocol = self::detectProtocol();
 247                 }
 248                 return $this->protocol;
 249         }
 250
 251         /**
 252          * Check for title, action, and/or variant data in the URL
 253          * and interpolate it into the GET variables.
 254          * This should only be run after $wgContLang is available,
 255          * as we may need the list of language variants to determine
 256          * available variant URLs.
 257          */
 258         public function interpolateTitle() {
 259                 // bug 16019: title interpolation on API queries is useless and sometimes harmful
 260                 if ( defined( 'MW_API' ) ) {
 261                         return;
 262                 }
 263
 264                 $matches = self::getPathInfo( 'title' );
 265                 foreach ( $matches as $key => $val ) {
 266                         $this->data[$key] = $_GET[$key] = $_REQUEST[$key] = $val;
 267                 }
 268         }
 269
 270         /**
 271          * URL rewriting function; tries to extract page title and,
 272          * optionally, one other fixed parameter value from a URL path.
 273          *
 274          * @param string $path The URL path given from the client
 275          * @param array $bases One or more URLs, optionally with $1 at the end
 276          * @param string $key If provided, the matching key in $bases will be
 277          *    passed on as the value of this URL parameter
 278          * @return array Array of URL variables to interpolate; empty if no match
 279          */
 280         static function extractTitle( $path, $bases, $key = false ) {
 281                 foreach ( (array)$bases as $keyValue => $base ) {
 282                         // Find the part after $wgArticlePath
 283                         $base = str_replace( '$1', '', $base );
 284                         $baseLen = strlen( $base );
 285                         if ( substr( $path, 0, $baseLen ) == $base ) {
 286                                 $raw = substr( $path, $baseLen );
 287                                 if ( $raw !== '' ) {
 288                                         $matches = array( 'title' => rawurldecode( $raw ) );
 289                                         if ( $key ) {
 290                                                 $matches[$key] = $keyValue;
 291                                         }
 292                                         return $matches;
 293                                 }
 294                         }
 295                 }
 296                 return array();
 297         }
 298
 299         /**
 300          * Recursively normalizes UTF-8 strings in the given array.
 301          *
 302          * @param string|array $data
 303          * @return array|string Cleaned-up version of the given
 304          * @private
 305          */
 306         function normalizeUnicode( $data ) {
 307                 if ( is_array( $data ) ) {
 308                         foreach ( $data as $key => $val ) {
 309                                 $data[$key] = $this->normalizeUnicode( $val );
 310                         }
 311                 } else {
 312                         global $wgContLang;
 313                         $data = isset( $wgContLang ) ?
 314                                 $wgContLang->normalize( $data ) :
 315                                 UtfNormal\Validator::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 array $arr
 324          * @param string $name
 325          * @param mixed $default
 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                         return $default;
 346                 }
 347         }
 348
 349         /**
 350          * Fetch a scalar from the input or return $default if it's not set.
 351          * Returns a string. Arrays are discarded. Useful for
 352          * non-freeform text inputs (e.g. predefined internal text keys
 353          * selected by a drop-down menu). For freeform input, see getText().
 354          *
 355          * @param string $name
 356          * @param string $default Optional default (or null)
 357          * @return string
 358          */
 359         public function getVal( $name, $default = null ) {
 360                 $val = $this->getGPCVal( $this->data, $name, $default );
 361                 if ( is_array( $val ) ) {
 362                         $val = $default;
 363                 }
 364                 if ( is_null( $val ) ) {
 365                         return $val;
 366                 } else {
 367                         return (string)$val;
 368                 }
 369         }
 370
 371         /**
 372          * Set an arbitrary value into our get/post data.
 373          *
 374          * @param string $key Key name to use
 375          * @param mixed $value Value to set
 376          * @return mixed Old value if one was present, null otherwise
 377          */
 378         public function setVal( $key, $value ) {
 379                 $ret = isset( $this->data[$key] ) ? $this->data[$key] : null;
 380                 $this->data[$key] = $value;
 381                 return $ret;
 382         }
 383
 384         /**
 385          * Unset an arbitrary value from our get/post data.
 386          *
 387          * @param string $key Key name to use
 388          * @return mixed Old value if one was present, null otherwise
 389          */
 390         public function unsetVal( $key ) {
 391                 if ( !isset( $this->data[$key] ) ) {
 392                         $ret = null;
 393                 } else {
 394                         $ret = $this->data[$key];
 395                         unset( $this->data[$key] );
 396                 }
 397                 return $ret;
 398         }
 399
 400         /**
 401          * Fetch an array from the input or return $default if it's not set.
 402          * If source was scalar, will return an array with a single element.
 403          * If no source and no default, returns null.
 404          *
 405          * @param string $name
 406          * @param array $default Optional default (or null)
 407          * @return array
 408          */
 409         public function getArray( $name, $default = null ) {
 410                 $val = $this->getGPCVal( $this->data, $name, $default );
 411                 if ( is_null( $val ) ) {
 412                         return null;
 413                 } else {
 414                         return (array)$val;
 415                 }
 416         }
 417
 418         /**
 419          * Fetch an array of integers, or return $default if it's not set.
 420          * If source was scalar, will return an array with a single element.
 421          * If no source and no default, returns null.
 422          * If an array is returned, contents are guaranteed to be integers.
 423          *
 424          * @param string $name
 425          * @param array $default Option default (or null)
 426          * @return array Array of ints
 427          */
 428         public function getIntArray( $name, $default = null ) {
 429                 $val = $this->getArray( $name, $default );
 430                 if ( is_array( $val ) ) {
 431                         $val = array_map( 'intval', $val );
 432                 }
 433                 return $val;
 434         }
 435
 436         /**
 437          * Fetch an integer value from the input or return $default if not set.
 438          * Guaranteed to return an integer; non-numeric input will typically
 439          * return 0.
 440          *
 441          * @param string $name
 442          * @param int $default
 443          * @return int
 444          */
 445         public function getInt( $name, $default = 0 ) {
 446                 return intval( $this->getVal( $name, $default ) );
 447         }
 448
 449         /**
 450          * Fetch an integer value from the input or return null if empty.
 451          * Guaranteed to return an integer or null; non-numeric input will
 452          * typically return null.
 453          *
 454          * @param string $name
 455          * @return int|null
 456          */
 457         public function getIntOrNull( $name ) {
 458                 $val = $this->getVal( $name );
 459                 return is_numeric( $val )
 460                         ? intval( $val )
 461                         : null;
 462         }
 463
 464         /**
 465          * Fetch a floating point value from the input or return $default if not set.
 466          * Guaranteed to return a float; non-numeric input will typically
 467          * return 0.
 468          *
 469          * @since 1.23
 470          * @param string $name
 471          * @param float $default
 472          * @return float
 473          */
 474         public function getFloat( $name, $default = 0.0 ) {
 475                 return floatval( $this->getVal( $name, $default ) );
 476         }
 477
 478         /**
 479          * Fetch a boolean value from the input or return $default if not set.
 480          * Guaranteed to return true or false, with normal PHP semantics for
 481          * boolean interpretation of strings.
 482          *
 483          * @param string $name
 484          * @param bool $default
 485          * @return bool
 486          */
 487         public function getBool( $name, $default = false ) {
 488                 return (bool)$this->getVal( $name, $default );
 489         }
 490
 491         /**
 492          * Fetch a boolean value from the input or return $default if not set.
 493          * Unlike getBool, the string "false" will result in boolean false, which is
 494          * useful when interpreting information sent from JavaScript.
 495          *
 496          * @param string $name
 497          * @param bool $default
 498          * @return bool
 499          */
 500         public function getFuzzyBool( $name, $default = false ) {
 501                 return $this->getBool( $name, $default ) && strcasecmp( $this->getVal( $name ), 'false' ) !== 0;
 502         }
 503
 504         /**
 505          * Return true if the named value is set in the input, whatever that
 506          * value is (even "0"). Return false if the named value is not set.
 507          * Example use is checking for the presence of check boxes in forms.
 508          *
 509          * @param string $name
 510          * @return bool
 511          */
 512         public function getCheck( $name ) {
 513                 # Checkboxes and buttons are only present when clicked
 514                 # Presence connotes truth, absence false
 515                 return $this->getVal( $name, null ) !== null;
 516         }
 517
 518         /**
 519          * Fetch a text string from the given array or return $default if it's not
 520          * set. Carriage returns are stripped from the text, and with some language
 521          * modules there is an input transliteration applied. This should generally
 522          * be used for form "<textarea>" and "<input>" fields. Used for
 523          * user-supplied freeform text input (for which input transformations may
 524          * be required - e.g.  Esperanto x-coding).
 525          *
 526          * @param string $name
 527          * @param string $default Optional
 528          * @return string
 529          */
 530         public function getText( $name, $default = '' ) {
 531                 global $wgContLang;
 532                 $val = $this->getVal( $name, $default );
 533                 return str_replace( "\r\n", "\n",
 534                         $wgContLang->recodeInput( $val ) );
 535         }
 536
 537         /**
 538          * Extracts the given named values into an array.
 539          * If no arguments are given, returns all input values.
 540          * No transformation is performed on the values.
 541          *
 542          * @return array
 543          */
 544         public function getValues() {
 545                 $names = func_get_args();
 546                 if ( count( $names ) == 0 ) {
 547                         $names = array_keys( $this->data );
 548                 }
 549
 550                 $retVal = array();
 551                 foreach ( $names as $name ) {
 552                         $value = $this->getGPCVal( $this->data, $name, null );
 553                         if ( !is_null( $value ) ) {
 554                                 $retVal[$name] = $value;
 555                         }
 556                 }
 557                 return $retVal;
 558         }
 559
 560         /**
 561          * Returns the names of all input values excluding those in $exclude.
 562          *
 563          * @param array $exclude
 564          * @return array
 565          */
 566         public function getValueNames( $exclude = array() ) {
 567                 return array_diff( array_keys( $this->getValues() ), $exclude );
 568         }
 569
 570         /**
 571          * Get the values passed in the query string.
 572          * No transformation is performed on the values.
 573          *
 574          * @return array
 575          */
 576         public function getQueryValues() {
 577                 return $_GET;
 578         }
 579
 580         /**
 581          * Return the contents of the Query with no decoding. Use when you need to
 582          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 583          *
 584          * @return string
 585          */
 586         public function getRawQueryString() {
 587                 return $_SERVER['QUERY_STRING'];
 588         }
 589
 590         /**
 591          * Return the contents of the POST with no decoding. Use when you need to
 592          * know exactly what was sent, e.g. for an OAuth signature over the elements.
 593          *
 594          * @return string
 595          */
 596         public function getRawPostString() {
 597                 if ( !$this->wasPosted() ) {
 598                         return '';
 599                 }
 600                 return $this->getRawInput();
 601         }
 602
 603         /**
 604          * Return the raw request body, with no processing. Cached since some methods
 605          * disallow reading the stream more than once. As stated in the php docs, this
 606          * does not work with enctype="multipart/form-data".
 607          *
 608          * @return string
 609          */
 610         public function getRawInput() {
 611                 static $input = null;
 612                 if ( $input === null ) {
 613                         $input = file_get_contents( 'php://input' );
 614                 }
 615                 return $input;
 616         }
 617
 618         /**
 619          * Get the HTTP method used for this request.
 620          *
 621          * @return string
 622          */
 623         public function getMethod() {
 624                 return isset( $_SERVER['REQUEST_METHOD'] ) ? $_SERVER['REQUEST_METHOD'] : 'GET';
 625         }
 626
 627         /**
 628          * Returns true if the present request was reached by a POST operation,
 629          * false otherwise (GET, HEAD, or command-line).
 630          *
 631          * Note that values retrieved by the object may come from the
 632          * GET URL etc even on a POST request.
 633          *
 634          * @return bool
 635          */
 636         public function wasPosted() {
 637                 return $this->getMethod() == 'POST';
 638         }
 639
 640         /**
 641          * Returns true if there is a session cookie set.
 642          * This does not necessarily mean that the user is logged in!
 643          *
 644          * If you want to check for an open session, use session_id()
 645          * instead; that will also tell you if the session was opened
 646          * during the current request (in which case the cookie will
 647          * be sent back to the client at the end of the script run).
 648          *
 649          * @return bool
 650          */
 651         public function checkSessionCookie() {
 652                 return isset( $_COOKIE[session_name()] );
 653         }
 654
 655         /**
 656          * Get a cookie from the $_COOKIE jar
 657          *
 658          * @param string $key The name of the cookie
 659          * @param string $prefix A prefix to use for the cookie name, if not $wgCookiePrefix
 660          * @param mixed $default What to return if the value isn't found
 661          * @return mixed Cookie value or $default if the cookie not set
 662          */
 663         public function getCookie( $key, $prefix = null, $default = null ) {
 664                 if ( $prefix === null ) {
 665                         global $wgCookiePrefix;
 666                         $prefix = $wgCookiePrefix;
 667                 }
 668                 return $this->getGPCVal( $_COOKIE, $prefix . $key, $default );
 669         }
 670
 671         /**
 672          * Return the path and query string portion of the request URI.
 673          * This will be suitable for use as a relative link in HTML output.
 674          *
 675          * @throws MWException
 676          * @return string
 677          */
 678         public function getRequestURL() {
 679                 if ( isset( $_SERVER['REQUEST_URI'] ) && strlen( $_SERVER['REQUEST_URI'] ) ) {
 680                         $base = $_SERVER['REQUEST_URI'];
 681                 } elseif ( isset( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 682                         && strlen( $_SERVER['HTTP_X_ORIGINAL_URL'] )
 683                 ) {
 684                         // Probably IIS; doesn't set REQUEST_URI
 685                         $base = $_SERVER['HTTP_X_ORIGINAL_URL'];
 686                 } elseif ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
 687                         $base = $_SERVER['SCRIPT_NAME'];
 688                         if ( isset( $_SERVER['QUERY_STRING'] ) && $_SERVER['QUERY_STRING'] != '' ) {
 689                                 $base .= '?' . $_SERVER['QUERY_STRING'];
 690                         }
 691                 } else {
 692                         // This shouldn't happen!
 693                         throw new MWException( "Web server doesn't provide either " .
 694                                 "REQUEST_URI, HTTP_X_ORIGINAL_URL or SCRIPT_NAME. Report details " .
 695                                 "of your web server configuration to https://phabricator.wikimedia.org/" );
 696                 }
 697                 // User-agents should not send a fragment with the URI, but
 698                 // if they do, and the web server passes it on to us, we
 699                 // need to strip it or we get false-positive redirect loops
 700                 // or weird output URLs
 701                 $hash = strpos( $base, '#' );
 702                 if ( $hash !== false ) {
 703                         $base = substr( $base, 0, $hash );
 704                 }
 705
 706                 if ( $base[0] == '/' ) {
 707                         // More than one slash will look like it is protocol relative
 708                         return preg_replace( '!^/+!', '/', $base );
 709                 } else {
 710                         // We may get paths with a host prepended; strip it.
 711                         return preg_replace( '!^[^:]+://[^/]+/+!', '/', $base );
 712                 }
 713         }
 714
 715         /**
 716          * Return the request URI with the canonical service and hostname, path,
 717          * and query string. This will be suitable for use as an absolute link
 718          * in HTML or other output.
 719          *
 720          * If $wgServer is protocol-relative, this will return a fully
 721          * qualified URL with the protocol that was used for this request.
 722          *
 723          * @return string
 724          */
 725         public function getFullRequestURL() {
 726                 return wfExpandUrl( $this->getRequestURL(), PROTO_CURRENT );
 727         }
 728
 729         /**
 730          * @param string $key
 731          * @param string $value
 732          * @return string
 733          */
 734         public function appendQueryValue( $key, $value ) {
 735                 return $this->appendQueryArray( array( $key => $value ) );
 736         }
 737
 738         /**
 739          * Appends or replaces value of query variables.
 740          *
 741          * @param array $array Array of values to replace/add to query
 742          * @return string
 743          */
 744         public function appendQueryArray( $array ) {
 745                 $newquery = $this->getQueryValues();
 746                 unset( $newquery['title'] );
 747                 $newquery = array_merge( $newquery, $array );
 748
 749                 return wfArrayToCgi( $newquery );
 750         }
 751
 752         /**
 753          * Check for limit and offset parameters on the input, and return sensible
 754          * defaults if not given. The limit must be positive and is capped at 5000.
 755          * Offset must be positive but is not capped.
 756          *
 757          * @param int $deflimit Limit to use if no input and the user hasn't set the option.
 758          * @param string $optionname To specify an option other than rclimit to pull from.
 759          * @return int[] First element is limit, second is offset
 760          */
 761         public function getLimitOffset( $deflimit = 50, $optionname = 'rclimit' ) {
 762                 global $wgUser;
 763
 764                 $limit = $this->getInt( 'limit', 0 );
 765                 if ( $limit < 0 ) {
 766                         $limit = 0;
 767                 }
 768                 if ( ( $limit == 0 ) && ( $optionname != '' ) ) {
 769                         $limit = $wgUser->getIntOption( $optionname );
 770                 }
 771                 if ( $limit <= 0 ) {
 772                         $limit = $deflimit;
 773                 }
 774                 if ( $limit > 5000 ) {
 775                         $limit = 5000; # We have *some* limits...
 776                 }
 777
 778                 $offset = $this->getInt( 'offset', 0 );
 779                 if ( $offset < 0 ) {
 780                         $offset = 0;
 781                 }
 782
 783                 return array( $limit, $offset );
 784         }
 785
 786         /**
 787          * Return the path to the temporary file where PHP has stored the upload.
 788          *
 789          * @param string $key
 790          * @return string|null String or null if no such file.
 791          */
 792         public function getFileTempname( $key ) {
 793                 $file = new WebRequestUpload( $this, $key );
 794                 return $file->getTempName();
 795         }
 796
 797         /**
 798          * Return the upload error or 0
 799          *
 800          * @param string $key
 801          * @return int
 802          */
 803         public function getUploadError( $key ) {
 804                 $file = new WebRequestUpload( $this, $key );
 805                 return $file->getError();
 806         }
 807
 808         /**
 809          * Return the original filename of the uploaded file, as reported by
 810          * the submitting user agent. HTML-style character entities are
 811          * interpreted and normalized to Unicode normalization form C, in part
 812          * to deal with weird input from Safari with non-ASCII filenames.
 813          *
 814          * Other than this the name is not verified for being a safe filename.
 815          *
 816          * @param string $key
 817          * @return string|null String or null if no such file.
 818          */
 819         public function getFileName( $key ) {
 820                 $file = new WebRequestUpload( $this, $key );
 821                 return $file->getName();
 822         }
 823
 824         /**
 825          * Return a WebRequestUpload object corresponding to the key
 826          *
 827          * @param string $key
 828          * @return WebRequestUpload
 829          */
 830         public function getUpload( $key ) {
 831                 return new WebRequestUpload( $this, $key );
 832         }
 833
 834         /**
 835          * Return a handle to WebResponse style object, for setting cookies,
 836          * headers and other stuff, for Request being worked on.
 837          *
 838          * @return WebResponse
 839          */
 840         public function response() {
 841                 /* Lazy initialization of response object for this request */
 842                 if ( !is_object( $this->response ) ) {
 843                         $class = ( $this instanceof FauxRequest ) ? 'FauxResponse' : 'WebResponse';
 844                         $this->response = new $class();
 845                 }
 846                 return $this->response;
 847         }
 848
 849         /**
 850          * Initialise the header list
 851          */
 852         protected function initHeaders() {
 853                 if ( count( $this->headers ) ) {
 854                         return;
 855                 }
 856
 857                 $apacheHeaders = function_exists( 'apache_request_headers' ) ? apache_request_headers() : false;
 858                 if ( $apacheHeaders ) {
 859                         foreach ( $apacheHeaders as $tempName => $tempValue ) {
 860                                 $this->headers[strtoupper( $tempName )] = $tempValue;
 861                         }
 862                 } else {
 863                         foreach ( $_SERVER as $name => $value ) {
 864                                 if ( substr( $name, 0, 5 ) === 'HTTP_' ) {
 865                                         $name = str_replace( '_', '-', substr( $name, 5 ) );
 866                                         $this->headers[$name] = $value;
 867                                 } elseif ( $name === 'CONTENT_LENGTH' ) {
 868                                         $this->headers['CONTENT-LENGTH'] = $value;
 869                                 }
 870                         }
 871                 }
 872         }
 873
 874         /**
 875          * Get an array containing all request headers
 876          *
 877          * @return array Mapping header name to its value
 878          */
 879         public function getAllHeaders() {
 880                 $this->initHeaders();
 881                 return $this->headers;
 882         }
 883
 884         /**
 885          * Get a request header, or false if it isn't set.
 886          *
 887          * @param string $name Case-insensitive header name
 888          * @param int $flags Bitwise combination of:
 889          *   WebRequest::GETHEADER_LIST  Treat the header as a comma-separated list
 890          *                               of values, as described in RFC 2616 § 4.2.
 891          *                               (since 1.26).
 892          * @return string|array|bool False if header is unset; otherwise the
 893          *  header value(s) as either a string (the default) or an array, if
 894          *  WebRequest::GETHEADER_LIST flag was set.
 895          */
 896         public function getHeader( $name, $flags = 0 ) {
 897                 $this->initHeaders();
 898                 $name = strtoupper( $name );
 899                 if ( !isset( $this->headers[$name] ) ) {
 900                         return false;
 901                 }
 902                 $value = $this->headers[$name];
 903                 if ( $flags & self::GETHEADER_LIST ) {
 904                         $value = array_map( 'trim', explode( ',', $value ) );
 905                 }
 906                 return $value;
 907         }
 908
 909         /**
 910          * Get data from $_SESSION
 911          *
 912          * @param string $key Name of key in $_SESSION
 913          * @return mixed
 914          */
 915         public function getSessionData( $key ) {
 916                 if ( !isset( $_SESSION[$key] ) ) {
 917                         return null;
 918                 }
 919                 return $_SESSION[$key];
 920         }
 921
 922         /**
 923          * Set session data
 924          *
 925          * @param string $key Name of key in $_SESSION
 926          * @param mixed $data
 927          */
 928         public function setSessionData( $key, $data ) {
 929                 $_SESSION[$key] = $data;
 930         }
 931
 932         /**
 933          * Check if Internet Explorer will detect an incorrect cache extension in
 934          * PATH_INFO or QUERY_STRING. If the request can't be allowed, show an error
 935          * message or redirect to a safer URL. Returns true if the URL is OK, and
 936          * false if an error message has been shown and the request should be aborted.
 937          *
 938          * @param array $extWhitelist
 939          * @throws HttpError
 940          * @return bool
 941          */
 942         public function checkUrlExtension( $extWhitelist = array() ) {
 943                 $extWhitelist[] = 'php';
 944                 if ( IEUrlExtension::areServerVarsBad( $_SERVER, $extWhitelist ) ) {
 945                         if ( !$this->wasPosted() ) {
 946                                 $newUrl = IEUrlExtension::fixUrlForIE6(
 947                                         $this->getFullRequestURL(), $extWhitelist );
 948                                 if ( $newUrl !== false ) {
 949                                         $this->doSecurityRedirect( $newUrl );
 950                                         return false;
 951                                 }
 952                         }
 953                         throw new HttpError( 403,
 954                                 'Invalid file extension found in the path info or query string.' );
 955                 }
 956                 return true;
 957         }
 958
 959         /**
 960          * Attempt to redirect to a URL with a QUERY_STRING that's not dangerous in
 961          * IE 6. Returns true if it was successful, false otherwise.
 962          *
 963          * @param string $url
 964          * @return bool
 965          */
 966         protected function doSecurityRedirect( $url ) {
 967                 header( 'Location: ' . $url );
 968                 header( 'Content-Type: text/html' );
 969                 $encUrl = htmlspecialchars( $url );
 970                 echo <<<HTML
 971 <html>
 972 <head>
 973 <title>Security redirect</title>
 974 </head>
 975 <body>
 976 <h1>Security redirect</h1>
 977 <p>
 978 We can't serve non-HTML content from the URL you have requested, because
 979 Internet Explorer would interpret it as an incorrect and potentially dangerous
 980 content type.</p>
 981 <p>Instead, please use <a href="$encUrl">this URL</a>, which is the same as the
 982 URL you have requested, except that "&amp;*" is appended. This prevents Internet
 983 Explorer from seeing a bogus file extension.
 984 </p>
 985 </body>
 986 </html>
 987 HTML;
 988                 echo "\n";
 989                 return true;
 990         }
 991
 992         /**
 993          * Parse the Accept-Language header sent by the client into an array
 994          *
 995          * @return array Array( languageCode => q-value ) sorted by q-value in
 996          *   descending order then appearing time in the header in ascending order.
 997          * May contain the "language" '*', which applies to languages other than those explicitly listed.
 998          * This is aligned with rfc2616 section 14.4
 999          * Preference for earlier languages appears in rfc3282 as an extension to HTTP/1.1.
1000          */
1001         public function getAcceptLang() {
1002                 // Modified version of code found at
1003                 // http://www.thefutureoftheweb.com/blog/use-accept-language-header
1004                 $acceptLang = $this->getHeader( 'Accept-Language' );
1005                 if ( !$acceptLang ) {
1006                         return array();
1007                 }
1008
1009                 // Return the language codes in lower case
1010                 $acceptLang = strtolower( $acceptLang );
1011
1012                 // Break up string into pieces (languages and q factors)
1013                 $lang_parse = null;
1014                 preg_match_all(
1015                         '/([a-z]{1,8}(-[a-z]{1,8})*|\*)\s*(;\s*q\s*=\s*(1(\.0{0,3})?|0(\.[0-9]{0,3})?)?)?/',
1016                         $acceptLang,
1017                         $lang_parse
1018                 );
1019
1020                 if ( !count( $lang_parse[1] ) ) {
1021                         return array();
1022                 }
1023
1024                 $langcodes = $lang_parse[1];
1025                 $qvalues = $lang_parse[4];
1026                 $indices = range( 0, count( $lang_parse[1] ) - 1 );
1027
1028                 // Set default q factor to 1
1029                 foreach ( $indices as $index ) {
1030                         if ( $qvalues[$index] === '' ) {
1031                                 $qvalues[$index] = 1;
1032                         } elseif ( $qvalues[$index] == 0 ) {
1033                                 unset( $langcodes[$index], $qvalues[$index], $indices[$index] );
1034                         }
1035                 }
1036
1037                 // Sort list. First by $qvalues, then by order. Reorder $langcodes the same way
1038                 array_multisort( $qvalues, SORT_DESC, SORT_NUMERIC, $indices, $langcodes );
1039
1040                 // Create a list like "en" => 0.8
1041                 $langs = array_combine( $langcodes, $qvalues );
1042
1043                 return $langs;
1044         }
1045
1046         /**
1047          * Fetch the raw IP from the request
1048          *
1049          * @since 1.19
1050          *
1051          * @throws MWException
1052          * @return string
1053          */
1054         protected function getRawIP() {
1055                 if ( !isset( $_SERVER['REMOTE_ADDR'] ) ) {
1056                         return null;
1057                 }
1058
1059                 if ( is_array( $_SERVER['REMOTE_ADDR'] ) || strpos( $_SERVER['REMOTE_ADDR'], ',' ) !== false ) {
1060                         throw new MWException( __METHOD__
1061                                 . " : Could not determine the remote IP address due to multiple values." );
1062                 } else {
1063                         $ipchain = $_SERVER['REMOTE_ADDR'];
1064                 }
1065
1066                 return IP::canonicalize( $ipchain );
1067         }
1068
1069         /**
1070          * Work out the IP address based on various globals
1071          * For trusted proxies, use the XFF client IP (first of the chain)
1072          *
1073          * @since 1.19
1074          *
1075          * @throws MWException
1076          * @return string
1077          */
1078         public function getIP() {
1079                 global $wgUsePrivateIPs;
1080
1081                 # Return cached result
1082                 if ( $this->ip !== null ) {
1083                         return $this->ip;
1084                 }
1085
1086                 # collect the originating ips
1087                 $ip = $this->getRawIP();
1088                 if ( !$ip ) {
1089                         throw new MWException( 'Unable to determine IP.' );
1090                 }
1091
1092                 # Append XFF
1093                 $forwardedFor = $this->getHeader( 'X-Forwarded-For' );
1094                 if ( $forwardedFor !== false ) {
1095                         $isConfigured = IP::isConfiguredProxy( $ip );
1096                         $ipchain = array_map( 'trim', explode( ',', $forwardedFor ) );
1097                         $ipchain = array_reverse( $ipchain );
1098                         array_unshift( $ipchain, $ip );
1099
1100                         # Step through XFF list and find the last address in the list which is a
1101                         # trusted server. Set $ip to the IP address given by that trusted server,
1102                         # unless the address is not sensible (e.g. private). However, prefer private
1103                         # IP addresses over proxy servers controlled by this site (more sensible).
1104                         # Note that some XFF values might be "unknown" with Squid/Varnish.
1105                         foreach ( $ipchain as $i => $curIP ) {
1106                                 $curIP = IP::sanitizeIP( IP::canonicalize( $curIP ) );
1107                                 if ( !$curIP || !isset( $ipchain[$i + 1] ) || $ipchain[$i + 1] === 'unknown'
1108                                         || !IP::isTrustedProxy( $curIP )
1109                                 ) {
1110                                         break; // IP is not valid/trusted or does not point to anything
1111                                 }
1112                                 if (
1113                                         IP::isPublic( $ipchain[$i + 1] ) ||
1114                                         $wgUsePrivateIPs ||
1115                                         IP::isConfiguredProxy( $curIP ) // bug 48919; treat IP as sane
1116                                 ) {
1117                                         // Follow the next IP according to the proxy
1118                                         $nextIP = IP::canonicalize( $ipchain[$i + 1] );
1119                                         if ( !$nextIP && $isConfigured ) {
1120                                                 // We have not yet made it past CDN/proxy servers of this site,
1121                                                 // so either they are misconfigured or there is some IP spoofing.
1122                                                 throw new MWException( "Invalid IP given in XFF '$forwardedFor'." );
1123                                         }
1124                                         $ip = $nextIP;
1125                                         // keep traversing the chain
1126                                         continue;
1127                                 }
1128                                 break;
1129                         }
1130                 }
1131
1132                 # Allow extensions to improve our guess
1133                 Hooks::run( 'GetIP', array( &$ip ) );
1134
1135                 if ( !$ip ) {
1136                         throw new MWException( "Unable to determine IP." );
1137                 }
1138
1139                 wfDebug( "IP: $ip\n" );
1140                 $this->ip = $ip;
1141                 return $ip;
1142         }
1143
1144         /**
1145          * @param string $ip
1146          * @return void
1147          * @since 1.21
1148          */
1149         public function setIP( $ip ) {
1150                 $this->ip = $ip;
1151         }
1152 }