More rigorous fix for ImageMagick parameter interpretation (bug 23148 etc.) based...

[lhc/web/wiklou.git] / includes / MimeMagic.php

diff --git a/includes/MimeMagic.php b/includes/MimeMagic.php
index 8383c0d..13a2aa4 100644 (file)
--- a/includes/MimeMagic.php
+++ b/includes/MimeMagic.php
@@ -80,7 +80,7 @@ audio/x-aiff [AUDIO]
 audio/x-wav [AUDIO]
 audio/mp3 audio/mpeg [AUDIO]
 application/ogg audio/ogg video/ogg [MULTIMEDIA]
-image/x-bmp image/bmp [BITMAP]
+image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 image/gif [BITMAP]
 image/jpeg [BITMAP]
 image/png [BITMAP]
@@ -118,19 +118,23 @@ class MimeMagic {
        * Mapping of media types to arrays of mime types.
        * This is used by findMediaType and getMediaType, respectively
        */
-       var $mMediaTypes= NULL;
+       var $mMediaTypes= null;
 
        /** Map of mime type aliases
        */
-       var $mMimeTypeAliases= NULL;
+       var $mMimeTypeAliases= null;
 
        /** map of mime types to file extensions (as a space seprarated list)
        */
-       var $mMimeToExt= NULL;
+       var $mMimeToExt= null;
 
        /** map of file extensions types to mime types (as a space seprarated list)
        */
-       var $mExtToMime= NULL;
+       var $mExtToMime= null;
+
+       /** IEContentAnalyzer instance
+        */
+       var $mIEAnalyzer;
 
        /** The singleton instance
         */
@@ -324,7 +328,7 @@ class MimeMagic {
        */
        function guessTypesForExtension( $ext ) {
                $m = $this->getTypesForExtension( $ext );
-               if ( is_null( $m ) ) return NULL;
+               if ( is_null( $m ) ) return null;
 
                $m = trim( $m );
                $m = preg_replace( '/\s.*$/', '', $m );
@@ -341,7 +345,7 @@ class MimeMagic {
                $ext = $this->getExtensionsForType( $mime );
 
                if ( !$ext ) {
-                       return NULL;  //unknown
+                       return null;  //unknown
                }
 
                $ext = explode( ' ', $ext );
@@ -465,16 +469,18 @@ class MimeMagic {
                }
 
                /*
-                * look for PHP
-                * Check for this before HTML/XML...
-                * Warning: this is a heuristic, and won't match a file with a lot of non-PHP before.
-                * It will also match text files which could be PHP. :)
+                * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
+                * heuristic, and won't match a file with a lot of non-PHP before.  It
+                * will also match text files which could be PHP. :)
+                *
+                * FIXME: For this reason, the check is probably useless -- an attacker
+                * could almost certainly just pad the file with a lot of nonsense to
+                * circumvent the check in any case where it would be a security
+                * problem.  On the other hand, it causes harmful false positives (bug
+                * 16583).  The heuristic has been cut down to exclude three-character
+                * strings like "<? ", but should it be axed completely?
                 */
                if( ( strpos( $head, '<?php' ) !== false ) ||
-                   ( strpos( $head, '<? ' ) !== false ) ||
-                   ( strpos( $head, "<?\n" ) !== false ) ||
-                   ( strpos( $head, "<?\t" ) !== false ) ||
-                   ( strpos( $head, "<?=" ) !== false ) ||
 
                    ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
                    ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
@@ -502,7 +508,7 @@ class MimeMagic {
                /*
                 * look for shell scripts
                 */
-               $script_type = NULL;
+               $script_type = null;
 
                # detect by shebang
                if ( substr( $head, 0, 2) == "#!" ) {
@@ -575,22 +581,22 @@ class MimeMagic {
         */
        function detectZipType( $header ) {
                $opendocTypes = array(
-                       'chart',
                        'chart-template',
-                       'formula',
+                       'chart',
                        'formula-template',
-                       'graphics',
+                       'formula',
                        'graphics-template',
-                       'image',
+                       'graphics',
                        'image-template',
-                       'presentation',
+                       'image',
                        'presentation-template',
-                       'spreadsheet',
+                       'presentation',
                        'spreadsheet-template',
-                       'text',
+                       'spreadsheet',
                        'text-template',
                        'text-master',
-                       'text-web' );
+                       'text-web',
+                       'text' );
 
                // http://lists.oasis-open.org/archives/office/200505/msg00006.html
                $types = '(?:' . implode( '|', $opendocTypes ) . ')';
@@ -625,7 +631,7 @@ class MimeMagic {
        function detectMimeType( $file, $ext = true ) {
                global $wgMimeDetectorCommand;
 
-               $m = NULL;
+               $m = null;
                if ( $wgMimeDetectorCommand ) {
                        $fn = wfEscapeShellArg( $file );
                        $m = `$wgMimeDetectorCommand $fn`;
@@ -672,7 +678,7 @@ class MimeMagic {
                        $m = strtolower( $m );
 
                        if ( strpos( $m, 'unknown' ) !== false ) {
-                               $m = NULL;
+                               $m = null;
                        } else {
                                wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
                                return $m;
@@ -717,7 +723,7 @@ class MimeMagic {
        *
        * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
        */
-       function getMediaType( $path = NULL, $mime = NULL ) {
+       function getMediaType( $path = null, $mime = null ) {
                if( !$mime && !$path ) return MEDIATYPE_UNKNOWN;
 
                # If mime type is unknown, guess it
@@ -750,7 +756,7 @@ class MimeMagic {
                }
 
                # Check for entry for file extension
-               $e = NULL;
+               $e = null;
                if ( $path ) {
                        $i = strrpos( $path, '.' );
                        $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
@@ -808,4 +814,27 @@ class MimeMagic {
 
                return MEDIATYPE_UNKNOWN;
        }
+
+       /**
+        * Get the MIME types that various versions of Internet Explorer would 
+        * detect from a chunk of the content.
+        *
+        * @param string $fileName The file name (unused at present)
+        * @param string $chunk The first 256 bytes of the file
+        * @param string $proposed The MIME type proposed by the server
+        */
+       public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
+               $ca = $this->getIEContentAnalyzer();
+               return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
+       }
+
+       /**
+        * Get a cached instance of IEContentAnalyzer
+        */
+       protected function getIEContentAnalyzer() {
+               if ( is_null( $this->mIEAnalyzer ) ) {
+                       $this->mIEAnalyzer = new IEContentAnalyzer;
+               }
+               return $this->mIEAnalyzer;
+       }
 }