includes/RawPage.php

   1 <?php
   2 /**
   3  * Copyright (C) 2004 Gabriel Wicke <wicke@wikidev.net>
   4  * http://wikidev.net/
   5  * Based on PageHistory and SpecialExport
   6  *
   7  * License: GPL (http://www.gnu.org/copyleft/gpl.html)
   8  *
   9  * @author Gabriel Wicke <wicke@wikidev.net>
  10  */
  11
  12 /**
  13  * A simple method to retrieve the plain source of an article,
  14  * using "action=raw" in the GET request string.
  15  */
  16 class RawPage {
  17         var $mArticle, $mTitle, $mRequest;
  18         var $mOldId, $mGen, $mCharset;
  19         var $mSmaxage, $mMaxage;
  20         var $mContentType, $mExpandTemplates;
  21
  22         function __construct( &$article, $request = false ) {
  23                 global $wgRequest, $wgInputEncoding, $wgSquidMaxage, $wgJsMimeType;
  24
  25                 $allowedCTypes = array('text/x-wiki', $wgJsMimeType, 'text/css', 'application/x-zope-edit');
  26                 $this->mArticle =& $article;
  27                 $this->mTitle =& $article->mTitle;
  28
  29                 if ( $request === false ) {
  30                         $this->mRequest =& $wgRequest;
  31                 } else {
  32                         $this->mRequest = $request;
  33                 }
  34
  35                 $ctype = $this->mRequest->getVal( 'ctype' );
  36                 $smaxage = $this->mRequest->getIntOrNull( 'smaxage', $wgSquidMaxage );
  37                 $maxage = $this->mRequest->getInt( 'maxage', $wgSquidMaxage );
  38                 $this->mExpandTemplates = $this->mRequest->getVal( 'templates' ) === 'expand';
  39                 $this->mUseMessageCache = $this->mRequest->getBool( 'usemsgcache' );
  40
  41                 $oldid = $this->mRequest->getInt( 'oldid' );
  42                 switch ( $wgRequest->getText( 'direction' ) ) {
  43                         case 'next':
  44                                 # output next revision, or nothing if there isn't one
  45                                 if ( $oldid ) {
  46                                         $oldid = $this->mTitle->getNextRevisionId( $oldid );
  47                                 }
  48                                 $oldid = $oldid ? $oldid : -1;
  49                                 break;
  50                         case 'prev':
  51                                 # output previous revision, or nothing if there isn't one
  52                                 if ( ! $oldid ) {
  53                                         # get the current revision so we can get the penultimate one
  54                                         $this->mArticle->getTouched();
  55                                         $oldid = $this->mArticle->mLatest;
  56                                 }
  57                                 $prev = $this->mTitle->getPreviousRevisionId( $oldid );
  58                                 $oldid = $prev ? $prev : -1 ;
  59                                 break;
  60                         case 'cur':
  61                                 $oldid = 0;
  62                                 break;
  63                 }
  64                 $this->mOldId = $oldid;
  65
  66                 # special case for 'generated' raw things: user css/js
  67                 $gen = $this->mRequest->getVal( 'gen' );
  68
  69                 if($gen == 'css') {
  70                         $this->mGen = $gen;
  71                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  72                         if($ctype == '') $ctype = 'text/css';
  73                 } elseif ($gen == 'js') {
  74                         $this->mGen = $gen;
  75                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  76                         if($ctype == '') $ctype = $wgJsMimeType;
  77                 } else {
  78                         $this->mGen = false;
  79                 }
  80                 $this->mCharset = $wgInputEncoding;
  81                 $this->mSmaxage = intval( $smaxage );
  82                 $this->mMaxage = $maxage;
  83
  84                 // Output may contain user-specific data; vary for open sessions
  85                 $this->mPrivateCache = ( $this->mSmaxage == 0 ) ||
  86                         ( session_id() != '' );
  87
  88                 if ( $ctype == '' or ! in_array( $ctype, $allowedCTypes ) ) {
  89                         $this->mContentType = 'text/x-wiki';
  90                 } else {
  91                         $this->mContentType = $ctype;
  92                 }
  93         }
  94
  95         function view() {
  96                 global $wgOut, $wgScript;
  97
  98                 if( isset( $_SERVER['SCRIPT_URL'] ) ) {
  99                         # Normally we use PHP_SELF to get the URL to the script
 100                         # as it was called, minus the query string.
 101                         #
 102                         # Some sites use Apache rewrite rules to handle subdomains,
 103                         # and have PHP set up in a weird way that causes PHP_SELF
 104                         # to contain the rewritten URL instead of the one that the
 105                         # outside world sees.
 106                         #
 107                         # If in this mode, use SCRIPT_URL instead, which mod_rewrite
 108                         # provides containing the "before" URL.
 109                         $url = $_SERVER['SCRIPT_URL'];
 110                 } else {
 111                         $url = $_SERVER['PHP_SELF'];
 112                 }
 113
 114                 $ua = @$_SERVER['HTTP_USER_AGENT'];
 115                 if( strcmp( $wgScript, $url ) && strpos( $ua, 'MSIE' ) !== false ) {
 116                         # Internet Explorer will ignore the Content-Type header if it
 117                         # thinks it sees a file extension it recognizes. Make sure that
 118                         # all raw requests are done through the script node, which will
 119                         # have eg '.php' and should remain safe.
 120                         #
 121                         # We used to redirect to a canonical-form URL as a general
 122                         # backwards-compatibility / good-citizen nice thing. However
 123                         # a lot of servers are set up in buggy ways, resulting in
 124                         # redirect loops which hang the browser until the CSS load
 125                         # times out.
 126                         #
 127                         # Just return a 403 Forbidden and get it over with.
 128                         wfHttpError( 403, 'Forbidden',
 129                                 'Raw pages must be accessed through the primary script entry point.' );
 130                         return;
 131                 }
 132
 133                 header( "Content-type: ".$this->mContentType.'; charset='.$this->mCharset );
 134                 # allow the client to cache this for 24 hours
 135                 $mode = $this->mPrivateCache ? 'private' : 'public';
 136                 header( 'Cache-Control: '.$mode.', s-maxage='.$this->mSmaxage.', max-age='.$this->mMaxage );
 137                 $text = $this->getRawText();
 138
 139                 if( !wfRunHooks( 'RawPageViewBeforeOutput', array( &$this, &$text ) ) ) {
 140                         wfDebug( __METHOD__ . ': RawPageViewBeforeOutput hook broke raw page output.' );
 141                 }
 142
 143                 echo $text;
 144                 $wgOut->disable();
 145         }
 146
 147         function getRawText() {
 148                 global $wgUser, $wgOut, $wgRequest, $wgJsMimeType;
 149
 150                 /* Disable retrieving content pages as raw js/css */
 151                 $dangerousTypes = array( $wgJsMimeType, 'text/css' );
 152                 if ( in_array( $this->mContentType, $dangerousTypes ) &&
 153                                 !($this->mTitle->isCssOrJsPage() || $this->mTitle->isCssJsSubpage() ) )
 154                         return '/* EMPTY */';
 155
 156                 if($this->mGen) {
 157                         $sk = $wgUser->getSkin();
 158                         $sk->initPage($wgOut);
 159                         if($this->mGen == 'css') {
 160                                 return $sk->getUserStylesheet();
 161                         } else if($this->mGen == 'js') {
 162                                 return $sk->getUserJs();
 163                         }
 164                 } else {
 165                         return $this->getArticleText();
 166                 }
 167         }
 168
 169         function getArticleText() {
 170                 $found = false;
 171                 $text = '';
 172                 if( $this->mTitle ) {
 173                         // If it's a MediaWiki message we can just hit the message cache
 174                         if ( $this->mUseMessageCache && $this->mTitle->getNamespace() == NS_MEDIAWIKI ) {
 175                                 $key = $this->mTitle->getDBkey();
 176                                 $text = wfMsgForContentNoTrans( $key );
 177                                 # If the message doesn't exist, return a blank
 178                                 if( wfEmptyMsg( $key, $text ) )
 179                                         $text = '';
 180                                 $found = true;
 181                         } else {
 182                                 // Get it from the DB
 183                                 $rev = Revision::newFromTitle( $this->mTitle, $this->mOldId );
 184                                 if ( $rev ) {
 185                                         $lastmod = wfTimestamp( TS_RFC2822, $rev->getTimestamp() );
 186                                         header( "Last-modified: $lastmod" );
 187                                         $text = $rev->getText();
 188                                         $found = true;
 189                                 }
 190                         }
 191                 }
 192
 193                 # Bad title or page does not exist
 194                 if( !$found && $this->mContentType == 'text/x-wiki' ) {
 195                         # Don't return a 404 response for CSS or JavaScript;
 196                         # 404s aren't generally cached and it would create
 197                         # extra hits when user CSS/JS are on and the user doesn't
 198                         # have the pages.
 199                         header( "HTTP/1.0 404 Not Found" );
 200                 }
 201
 202                 // Special-case for empty CSS/JS
 203                 //
 204                 // Internet Explorer for Mac handles empty files badly;
 205                 // particularly so when keep-alive is active. It can lead
 206                 // to long timeouts as it seems to sit there waiting for
 207                 // more data that never comes.
 208                 //
 209                 // Give it a comment...
 210                 if( strlen( $text ) == 0 &&
 211                         ($this->mContentType == 'text/css' ||
 212                                 $this->mContentType == 'text/javascript' ) ) {
 213                         return "/* Empty */";
 214                 }
 215
 216                 return $this->parseArticleText( $text );
 217         }
 218
 219         function parseArticleText( $text ) {
 220                 if ( $text === '' )
 221                         return '';
 222                 else
 223                         if ( $this->mExpandTemplates ) {
 224                                 global $wgParser;
 225                                 return $wgParser->preprocess( $text, $this->mTitle, new ParserOptions() );
 226                         } else
 227                                 return $text;
 228         }
 229 }
 230