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, $mSection;
  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                 $this->mSection = $this->mRequest->getIntOrNull( 'section' );
  42
  43                 $oldid = $this->mRequest->getInt( 'oldid' );
  44
  45                 switch ( $wgRequest->getText( 'direction' ) ) {
  46                         case 'next':
  47                                 # output next revision, or nothing if there isn't one
  48                                 if ( $oldid ) {
  49                                         $oldid = $this->mTitle->getNextRevisionId( $oldid );
  50                                 }
  51                                 $oldid = $oldid ? $oldid : -1;
  52                                 break;
  53                         case 'prev':
  54                                 # output previous revision, or nothing if there isn't one
  55                                 if ( ! $oldid ) {
  56                                         # get the current revision so we can get the penultimate one
  57                                         $this->mArticle->getTouched();
  58                                         $oldid = $this->mArticle->mLatest;
  59                                 }
  60                                 $prev = $this->mTitle->getPreviousRevisionId( $oldid );
  61                                 $oldid = $prev ? $prev : -1 ;
  62                                 break;
  63                         case 'cur':
  64                                 $oldid = 0;
  65                                 break;
  66                 }
  67                 $this->mOldId = $oldid;
  68
  69                 # special case for 'generated' raw things: user css/js
  70                 $gen = $this->mRequest->getVal( 'gen' );
  71
  72                 if($gen == 'css') {
  73                         $this->mGen = $gen;
  74                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  75                         if($ctype == '') $ctype = 'text/css';
  76                 } elseif ($gen == 'js') {
  77                         $this->mGen = $gen;
  78                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  79                         if($ctype == '') $ctype = $wgJsMimeType;
  80                 } else {
  81                         $this->mGen = false;
  82                 }
  83                 $this->mCharset = $wgInputEncoding;
  84                 $this->mSmaxage = intval( $smaxage );
  85                 $this->mMaxage = $maxage;
  86
  87                 // Output may contain user-specific data; vary for open sessions
  88                 $this->mPrivateCache = ( $this->mSmaxage == 0 ) ||
  89                         ( session_id() != '' );
  90
  91                 if ( $ctype == '' or ! in_array( $ctype, $allowedCTypes ) ) {
  92                         $this->mContentType = 'text/x-wiki';
  93                 } else {
  94                         $this->mContentType = $ctype;
  95                 }
  96         }
  97
  98         function view() {
  99                 global $wgOut, $wgScript;
 100
 101                 if( isset( $_SERVER['SCRIPT_URL'] ) ) {
 102                         # Normally we use PHP_SELF to get the URL to the script
 103                         # as it was called, minus the query string.
 104                         #
 105                         # Some sites use Apache rewrite rules to handle subdomains,
 106                         # and have PHP set up in a weird way that causes PHP_SELF
 107                         # to contain the rewritten URL instead of the one that the
 108                         # outside world sees.
 109                         #
 110                         # If in this mode, use SCRIPT_URL instead, which mod_rewrite
 111                         # provides containing the "before" URL.
 112                         $url = $_SERVER['SCRIPT_URL'];
 113                 } else {
 114                         $url = $_SERVER['PHP_SELF'];
 115                 }
 116
 117                 $ua = @$_SERVER['HTTP_USER_AGENT'];
 118                 if( strcmp( $wgScript, $url ) && strpos( $ua, 'MSIE' ) !== false ) {
 119                         # Internet Explorer will ignore the Content-Type header if it
 120                         # thinks it sees a file extension it recognizes. Make sure that
 121                         # all raw requests are done through the script node, which will
 122                         # have eg '.php' and should remain safe.
 123                         #
 124                         # We used to redirect to a canonical-form URL as a general
 125                         # backwards-compatibility / good-citizen nice thing. However
 126                         # a lot of servers are set up in buggy ways, resulting in
 127                         # redirect loops which hang the browser until the CSS load
 128                         # times out.
 129                         #
 130                         # Just return a 403 Forbidden and get it over with.
 131                         wfHttpError( 403, 'Forbidden',
 132                                 'Raw pages must be accessed through the primary script entry point.' );
 133                         return;
 134                 }
 135
 136                 header( "Content-type: ".$this->mContentType.'; charset='.$this->mCharset );
 137                 # allow the client to cache this for 24 hours
 138                 $mode = $this->mPrivateCache ? 'private' : 'public';
 139                 header( 'Cache-Control: '.$mode.', s-maxage='.$this->mSmaxage.', max-age='.$this->mMaxage );
 140                 $text = $this->getRawText();
 141
 142                 if( !wfRunHooks( 'RawPageViewBeforeOutput', array( &$this, &$text ) ) ) {
 143                         wfDebug( __METHOD__ . ': RawPageViewBeforeOutput hook broke raw page output.' );
 144                 }
 145
 146                 echo $text;
 147                 $wgOut->disable();
 148         }
 149
 150         function getRawText() {
 151                 global $wgUser, $wgOut, $wgRequest;
 152                 if($this->mGen) {
 153                         $sk = $wgUser->getSkin();
 154                         $sk->initPage($wgOut);
 155                         if($this->mGen == 'css') {
 156                                 return $sk->getUserStylesheet();
 157                         } else if($this->mGen == 'js') {
 158                                 return $sk->getUserJs();
 159                         }
 160                 } else {
 161                         return $this->getArticleText();
 162                 }
 163         }
 164
 165         function getArticleText() {
 166                 $found = false;
 167                 $text = '';
 168                 if( $this->mTitle ) {
 169                         // If it's a MediaWiki message we can just hit the message cache
 170                         if ( $this->mUseMessageCache && $this->mTitle->getNamespace() == NS_MEDIAWIKI ) {
 171                                 $key = $this->mTitle->getDBkey();
 172                                 $text = wfMsgForContentNoTrans( $key );
 173                                 # If the message doesn't exist, return a blank
 174                                 if( wfEmptyMsg( $key, $text ) )
 175                                         $text = '';
 176                                 $found = true;
 177                         } else {
 178                                 // Get it from the DB
 179                                 $rev = Revision::newFromTitle( $this->mTitle, $this->mOldId );
 180                                 if ( $rev ) {
 181                                         $lastmod = wfTimestamp( TS_RFC2822, $rev->getTimestamp() );
 182                                         header( "Last-modified: $lastmod" );
 183
 184                                         if ( !is_null($this->mSection) && $this->mSection != '' ) {
 185                                                 global $wgParser;
 186                                                 return $wgParser->getSection ( $rev->getText(), $this->mSection );
 187                                         } else
 188                                                 $text = $rev->getText();
 189                                         $found = true;
 190                                 }
 191                         }
 192                 }
 193
 194                 # Bad title or page does not exist
 195                 if( !$found && $this->mContentType == 'text/x-wiki' ) {
 196                         # Don't return a 404 response for CSS or JavaScript;
 197                         # 404s aren't generally cached and it would create
 198                         # extra hits when user CSS/JS are on and the user doesn't
 199                         # have the pages.
 200                         header( "HTTP/1.0 404 Not Found" );
 201                 }
 202
 203                 // Special-case for empty CSS/JS
 204                 //
 205                 // Internet Explorer for Mac handles empty files badly;
 206                 // particularly so when keep-alive is active. It can lead
 207                 // to long timeouts as it seems to sit there waiting for
 208                 // more data that never comes.
 209                 //
 210                 // Give it a comment...
 211                 if( strlen( $text ) == 0 &&
 212                         ($this->mContentType == 'text/css' ||
 213                                 $this->mContentType == 'text/javascript' ) ) {
 214                         return "/* Empty */";
 215                 }
 216
 217                 return $this->parseArticleText( $text );
 218         }
 219
 220         function parseArticleText( $text ) {
 221                 if ( $text === '' )
 222                         return '';
 223                 else
 224                         if ( $this->mExpandTemplates ) {
 225                                 global $wgParser;
 226                                 return $wgParser->preprocess( $text, $this->mTitle, new ParserOptions() );
 227                         } else
 228                                 return $text;
 229         }
 230 }
 231