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  * @file
  11  */
  12
  13 /**
  14  * A simple method to retrieve the plain source of an article,
  15  * using "action=raw" in the GET request string.
  16  */
  17 class RawPage {
  18         var $mArticle, $mTitle, $mRequest;
  19         var $mOldId, $mGen, $mCharset, $mSection;
  20         var $mSmaxage, $mMaxage;
  21         var $mContentType, $mExpandTemplates;
  22
  23         function __construct( &$article, $request = false ) {
  24                 global $wgRequest, $wgInputEncoding, $wgSquidMaxage, $wgJsMimeType, $wgGroupPermissions;
  25
  26                 $allowedCTypes = array('text/x-wiki', $wgJsMimeType, 'text/css', 'application/x-zope-edit');
  27                 $this->mArticle =& $article;
  28                 $this->mTitle =& $article->mTitle;
  29
  30                 if( $request === false ) {
  31                         $this->mRequest =& $wgRequest;
  32                 } else {
  33                         $this->mRequest = $request;
  34                 }
  35
  36                 $ctype = $this->mRequest->getVal( 'ctype' );
  37                 $smaxage = $this->mRequest->getIntOrNull( 'smaxage' );
  38                 $maxage = $this->mRequest->getInt( 'maxage', $wgSquidMaxage );
  39
  40                 $this->mExpandTemplates = $this->mRequest->getVal( 'templates' ) === 'expand';
  41                 $this->mUseMessageCache = $this->mRequest->getBool( 'usemsgcache' );
  42
  43                 $this->mSection = $this->mRequest->getIntOrNull( 'section' );
  44
  45                 $oldid = $this->mRequest->getInt( 'oldid' );
  46
  47                 switch( $wgRequest->getText( 'direction' ) ) {
  48                         case 'next':
  49                                 # output next revision, or nothing if there isn't one
  50                                 if( $oldid ) {
  51                                         $oldid = $this->mTitle->getNextRevisionId( $oldid );
  52                                 }
  53                                 $oldid = $oldid ? $oldid : -1;
  54                                 break;
  55                         case 'prev':
  56                                 # output previous revision, or nothing if there isn't one
  57                                 if( ! $oldid ) {
  58                                         # get the current revision so we can get the penultimate one
  59                                         $this->mArticle->getTouched();
  60                                         $oldid = $this->mArticle->mLatest;
  61                                 }
  62                                 $prev = $this->mTitle->getPreviousRevisionId( $oldid );
  63                                 $oldid = $prev ? $prev : -1 ;
  64                                 break;
  65                         case 'cur':
  66                                 $oldid = 0;
  67                                 break;
  68                 }
  69                 $this->mOldId = $oldid;
  70
  71                 # special case for 'generated' raw things: user css/js
  72                 $gen = $this->mRequest->getVal( 'gen' );
  73
  74                 if( $gen == 'css' ) {
  75                         $this->mGen = $gen;
  76                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  77                         if($ctype == '') $ctype = 'text/css';
  78                 } elseif( $gen == 'js' ) {
  79                         $this->mGen = $gen;
  80                         if( is_null( $smaxage ) ) $smaxage = $wgSquidMaxage;
  81                         if($ctype == '') $ctype = $wgJsMimeType;
  82                 } else {
  83                         $this->mGen = false;
  84                 }
  85                 $this->mCharset = $wgInputEncoding;
  86
  87                 # Force caching for CSS and JS raw content, default: 5 minutes
  88                 if( is_null($smaxage) and ($ctype=='text/css' or $ctype==$wgJsMimeType) ) {
  89                         global $wgForcedRawSMaxage;
  90                         $this->mSmaxage = intval($wgForcedRawSMaxage);
  91                 } else {
  92                         $this->mSmaxage = intval( $smaxage );
  93                 }
  94                 $this->mMaxage = $maxage;
  95
  96                 # Output may contain user-specific data;
  97                 # vary generated content for open sessions and private wikis
  98                 if( $this->mGen or !$wgGroupPermissions['*']['read'] ) {
  99                         $this->mPrivateCache = $this->mSmaxage == 0 || session_id() != '';
 100                 } else {
 101                         $this->mPrivateCache = false;
 102                 }
 103
 104                 if( $ctype == '' or ! in_array( $ctype, $allowedCTypes ) ) {
 105                         $this->mContentType = 'text/x-wiki';
 106                 } else {
 107                         $this->mContentType = $ctype;
 108                 }
 109         }
 110
 111         function view() {
 112                 global $wgOut, $wgScript;
 113
 114                 if( isset( $_SERVER['SCRIPT_URL'] ) ) {
 115                         # Normally we use PHP_SELF to get the URL to the script
 116                         # as it was called, minus the query string.
 117                         #
 118                         # Some sites use Apache rewrite rules to handle subdomains,
 119                         # and have PHP set up in a weird way that causes PHP_SELF
 120                         # to contain the rewritten URL instead of the one that the
 121                         # outside world sees.
 122                         #
 123                         # If in this mode, use SCRIPT_URL instead, which mod_rewrite
 124                         # provides containing the "before" URL.
 125                         $url = $_SERVER['SCRIPT_URL'];
 126                 } else {
 127                         $url = $_SERVER['PHP_SELF'];
 128                 }
 129
 130                 if( $url == '' ) {
 131                         # This will make the next check fail with a confusing error
 132                         # message, so we should mention it separately.
 133                         wfHttpError( 500, 'Internal Server Error',
 134                                 "\$_SERVER['PHP_SELF'] is not set.  Perhaps you're using CGI" .
 135                                 " and haven't set cgi.fix_pathinfo = 1 in php.ini?" );
 136                         return;
 137                 }
 138
 139                 if( strcmp( $wgScript, $url ) ) {
 140                         # Internet Explorer will ignore the Content-Type header if it
 141                         # thinks it sees a file extension it recognizes. Make sure that
 142                         # all raw requests are done through the script node, which will
 143                         # have eg '.php' and should remain safe.
 144                         #
 145                         # We used to redirect to a canonical-form URL as a general
 146                         # backwards-compatibility / good-citizen nice thing. However
 147                         # a lot of servers are set up in buggy ways, resulting in
 148                         # redirect loops which hang the browser until the CSS load
 149                         # times out.
 150                         #
 151                         # Just return a 403 Forbidden and get it over with.
 152                         wfHttpError( 403, 'Forbidden',
 153                                 'Raw pages must be accessed through the primary script entry point.' );
 154                         return;
 155                 }
 156
 157                 header( "Content-type: ".$this->mContentType.'; charset='.$this->mCharset );
 158                 # allow the client to cache this for 24 hours
 159                 $mode = $this->mPrivateCache ? 'private' : 'public';
 160                 header( 'Cache-Control: '.$mode.', s-maxage='.$this->mSmaxage.', max-age='.$this->mMaxage );
 161
 162                 if( HTMLFileCache::useFileCache() ) {
 163                         $cache = new HTMLFileCache( $this->mTitle, 'raw' );
 164                         if( $cache->isFileCacheGood( /* Assume up to date */ ) ) {
 165                                 $cache->loadFromFileCache();
 166                                 $wgOut->disable();
 167                                 return;
 168                         } else {
 169                                 ob_start( array(&$cache, 'saveToFileCache' ) );
 170                         }
 171                 }
 172
 173                 $text = $this->getRawText();
 174
 175                 if( !wfRunHooks( 'RawPageViewBeforeOutput', array( &$this, &$text ) ) ) {
 176                         wfDebug( __METHOD__ . ": RawPageViewBeforeOutput hook broke raw page output.\n" );
 177                 }
 178
 179                 echo $text;
 180                 $wgOut->disable();
 181         }
 182
 183         function getRawText() {
 184                 global $wgUser, $wgOut, $wgRequest;
 185                 if( $this->mGen ) {
 186                         $sk = $wgUser->getSkin();
 187                         if( !StubObject::isRealObject( $wgOut ) )
 188                                 $wgOut->_unstub( 2 );
 189                         $sk->initPage( $wgOut );
 190                         if( $this->mGen == 'css' ) {
 191                                 return $sk->generateUserStylesheet();
 192                         } else if( $this->mGen == 'js' ) {
 193                                 return $sk->generateUserJs();
 194                         }
 195                 } else {
 196                         return $this->getArticleText();
 197                 }
 198         }
 199
 200         function getArticleText() {
 201                 $found = false;
 202                 $text = '';
 203                 if( $this->mTitle ) {
 204                         // If it's a MediaWiki message we can just hit the message cache
 205                         if( $this->mUseMessageCache && $this->mTitle->getNamespace() == NS_MEDIAWIKI ) {
 206                                 $key = $this->mTitle->getDBkey();
 207                                 $text = wfMsgForContentNoTrans( $key );
 208                                 # If the message doesn't exist, return a blank
 209                                 if( wfEmptyMsg( $key, $text ) )
 210                                         $text = '';
 211                                 $found = true;
 212                         } else {
 213                                 // Get it from the DB
 214                                 $rev = Revision::newFromTitle( $this->mTitle, $this->mOldId );
 215                                 if( $rev ) {
 216                                         $lastmod = wfTimestamp( TS_RFC2822, $rev->getTimestamp() );
 217                                         header( "Last-modified: $lastmod" );
 218
 219                                         if( !is_null($this->mSection ) ) {
 220                                                 global $wgParser;
 221                                                 $text = $wgParser->getSection ( $rev->getText(), $this->mSection );
 222                                         } else
 223                                                 $text = $rev->getText();
 224                                         $found = true;
 225                                 }
 226                         }
 227                 }
 228
 229                 # Bad title or page does not exist
 230                 if( !$found && $this->mContentType == 'text/x-wiki' ) {
 231                         # Don't return a 404 response for CSS or JavaScript;
 232                         # 404s aren't generally cached and it would create
 233                         # extra hits when user CSS/JS are on and the user doesn't
 234                         # have the pages.
 235                         header( "HTTP/1.0 404 Not Found" );
 236                 }
 237
 238                 // Special-case for empty CSS/JS
 239                 //
 240                 // Internet Explorer for Mac handles empty files badly;
 241                 // particularly so when keep-alive is active. It can lead
 242                 // to long timeouts as it seems to sit there waiting for
 243                 // more data that never comes.
 244                 //
 245                 // Give it a comment...
 246                 if( strlen( $text ) == 0 &&
 247                         ($this->mContentType == 'text/css' ||
 248                                 $this->mContentType == 'text/javascript' ) ) {
 249                         return "/* Empty */";
 250                 }
 251
 252                 return $this->parseArticleText( $text );
 253         }
 254
 255         function parseArticleText( $text ) {
 256                 if( $text === '' )
 257                         return '';
 258                 else
 259                         if( $this->mExpandTemplates ) {
 260                                 global $wgParser;
 261                                 return $wgParser->preprocess( $text, $this->mTitle, new ParserOptions() );
 262                         } else
 263                                 return $text;
 264         }
 265 }