MimeAnalyzer.php

Go to the documentation of this file.

<?php
use Psr\Log\LoggerAwareInterface;
use Psr\Log\LoggerInterface;

class MimeAnalyzer implements LoggerAwareInterface {
    protected $typeFile;
    protected $infoFile;
    protected $xmlTypes;
    protected $initCallback;
    protected $detectCallback;
    protected $guessCallback;
    protected $extCallback;
    protected $mediaTypes = null;
    protected $mimeTypeAliases = null;
    protected $mimetoExt = null;

    public $mExtToMime = null; // legacy name; field accessed by hooks

    protected $IEAnalyzer;

    private $extraTypes = '';
    private $extraInfo = '';

    private $logger;

    protected static $wellKnownTypes = <<<EOT
application/ogg ogx ogg ogm ogv oga spx
application/pdf pdf
application/vnd.oasis.opendocument.chart odc
application/vnd.oasis.opendocument.chart-template otc
application/vnd.oasis.opendocument.database odb
application/vnd.oasis.opendocument.formula odf
application/vnd.oasis.opendocument.formula-template otf
application/vnd.oasis.opendocument.graphics odg
application/vnd.oasis.opendocument.graphics-template otg
application/vnd.oasis.opendocument.image odi
application/vnd.oasis.opendocument.image-template oti
application/vnd.oasis.opendocument.presentation odp
application/vnd.oasis.opendocument.presentation-template otp
application/vnd.oasis.opendocument.spreadsheet ods
application/vnd.oasis.opendocument.spreadsheet-template ots
application/vnd.oasis.opendocument.text odt
application/vnd.oasis.opendocument.text-master otm
application/vnd.oasis.opendocument.text-template ott
application/vnd.oasis.opendocument.text-web oth
application/javascript js
application/x-shockwave-flash swf
audio/midi mid midi kar
audio/mpeg mpga mpa mp2 mp3
audio/x-aiff aif aiff aifc
audio/x-wav wav
audio/ogg oga spx ogg
image/x-bmp bmp
image/gif gif
image/jpeg jpeg jpg jpe
image/png png
image/svg+xml svg
image/svg svg
image/tiff tiff tif
image/vnd.djvu djvu
image/x.djvu djvu
image/x-djvu djvu
image/x-portable-pixmap ppm
image/x-xcf xcf
text/plain txt
text/html html htm
video/ogg ogv ogm ogg
video/mpeg mpg mpeg
EOT;

    protected static $wellKnownInfo = <<<EOT
application/pdf [OFFICE]
application/vnd.oasis.opendocument.chart [OFFICE]
application/vnd.oasis.opendocument.chart-template [OFFICE]
application/vnd.oasis.opendocument.database [OFFICE]
application/vnd.oasis.opendocument.formula [OFFICE]
application/vnd.oasis.opendocument.formula-template [OFFICE]
application/vnd.oasis.opendocument.graphics [OFFICE]
application/vnd.oasis.opendocument.graphics-template [OFFICE]
application/vnd.oasis.opendocument.image [OFFICE]
application/vnd.oasis.opendocument.image-template [OFFICE]
application/vnd.oasis.opendocument.presentation [OFFICE]
application/vnd.oasis.opendocument.presentation-template [OFFICE]
application/vnd.oasis.opendocument.spreadsheet [OFFICE]
application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
application/vnd.oasis.opendocument.text [OFFICE]
application/vnd.oasis.opendocument.text-template [OFFICE]
application/vnd.oasis.opendocument.text-master [OFFICE]
application/vnd.oasis.opendocument.text-web [OFFICE]
application/javascript text/javascript application/x-javascript [EXECUTABLE]
application/x-shockwave-flash [MULTIMEDIA]
audio/midi [AUDIO]
audio/x-aiff [AUDIO]
audio/x-wav [AUDIO]
audio/mp3 audio/mpeg [AUDIO]
application/ogg audio/ogg video/ogg [MULTIMEDIA]
image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
image/gif [BITMAP]
image/jpeg [BITMAP]
image/png [BITMAP]
image/svg+xml [DRAWING]
image/tiff [BITMAP]
image/vnd.djvu [BITMAP]
image/x-xcf [BITMAP]
image/x-portable-pixmap [BITMAP]
text/plain [TEXT]
text/html [TEXT]
video/ogg [VIDEO]
video/mpeg [VIDEO]
unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
EOT;

    public function __construct( array $params ) {
        $this->typeFile = $params['typeFile'];
        $this->infoFile = $params['infoFile'];
        $this->xmlTypes = $params['xmlTypes'];
        $this->initCallback = isset( $params['initCallback'] )
            ? $params['initCallback']
            : null;
        $this->detectCallback = isset( $params['detectCallback'] )
            ? $params['detectCallback']
            : null;
        $this->guessCallback = isset( $params['guessCallback'] )
            ? $params['guessCallback']
            : null;
        $this->extCallback = isset( $params['extCallback'] )
            ? $params['extCallback']
            : null;
        $this->logger = isset( $params['logger'] )
            ? $params['logger']
            : new \Psr\Log\NullLogger();

        $this->loadFiles();
    }

    protected function loadFiles() {
        # Allow media handling extensions adding MIME-types and MIME-info
        if ( $this->initCallback ) {
            call_user_func( $this->initCallback, $this );
        }

        $types = self::$wellKnownTypes;

        $mimeTypeFile = $this->typeFile;
        if ( $mimeTypeFile ) {
            if ( is_file( $mimeTypeFile ) && is_readable( $mimeTypeFile ) ) {
                $this->logger->info( __METHOD__ . ": loading mime types from $mimeTypeFile\n" );
                $types .= "\n";
                $types .= file_get_contents( $mimeTypeFile );
            } else {
                $this->logger->info( __METHOD__ . ": can't load mime types from $mimeTypeFile\n" );
            }
        } else {
            $this->logger->info( __METHOD__ .
                ": no mime types file defined, using built-ins only.\n" );
        }

        $types .= "\n" . $this->extraTypes;

        $types = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $types );
        $types = str_replace( "\t", " ", $types );

        $this->mimetoExt = [];
        $this->mExtToMime = [];

        $lines = explode( "\n", $types );
        foreach ( $lines as $s ) {
            $s = trim( $s );
            if ( empty( $s ) ) {
                continue;
            }
            if ( strpos( $s, '#' ) === 0 ) {
                continue;
            }

            $s = strtolower( $s );
            $i = strpos( $s, ' ' );

            if ( $i === false ) {
                continue;
            }

            $mime = substr( $s, 0, $i );
            $ext = trim( substr( $s, $i + 1 ) );

            if ( empty( $ext ) ) {
                continue;
            }

            if ( !empty( $this->mimetoExt[$mime] ) ) {
                $this->mimetoExt[$mime] .= ' ' . $ext;
            } else {
                $this->mimetoExt[$mime] = $ext;
            }

            $extensions = explode( ' ', $ext );

            foreach ( $extensions as $e ) {
                $e = trim( $e );
                if ( empty( $e ) ) {
                    continue;
                }

                if ( !empty( $this->mExtToMime[$e] ) ) {
                    $this->mExtToMime[$e] .= ' ' . $mime;
                } else {
                    $this->mExtToMime[$e] = $mime;
                }
            }
        }

        $mimeInfoFile = $this->infoFile;

        $info = self::$wellKnownInfo;

        if ( $mimeInfoFile ) {
            if ( is_file( $mimeInfoFile ) && is_readable( $mimeInfoFile ) ) {
                $this->logger->info( __METHOD__ . ": loading mime info from $mimeInfoFile\n" );
                $info .= "\n";
                $info .= file_get_contents( $mimeInfoFile );
            } else {
                $this->logger->info( __METHOD__ . ": can't load mime info from $mimeInfoFile\n" );
            }
        } else {
            $this->logger->info( __METHOD__ .
                ": no mime info file defined, using built-ins only.\n" );
        }

        $info .= "\n" . $this->extraInfo;

        $info = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $info );
        $info = str_replace( "\t", " ", $info );

        $this->mimeTypeAliases = [];
        $this->mediaTypes = [];

        $lines = explode( "\n", $info );
        foreach ( $lines as $s ) {
            $s = trim( $s );
            if ( empty( $s ) ) {
                continue;
            }
            if ( strpos( $s, '#' ) === 0 ) {
                continue;
            }

            $s = strtolower( $s );
            $i = strpos( $s, ' ' );

            if ( $i === false ) {
                continue;
            }

            # print "processing MIME INFO line $s<br>";

            $match = [];
            if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
                $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
                $mtype = trim( strtoupper( $match[1] ) );
            } else {
                $mtype = MEDIATYPE_UNKNOWN;
            }

            $m = explode( ' ', $s );

            if ( !isset( $this->mediaTypes[$mtype] ) ) {
                $this->mediaTypes[$mtype] = [];
            }

            foreach ( $m as $mime ) {
                $mime = trim( $mime );
                if ( empty( $mime ) ) {
                    continue;
                }

                $this->mediaTypes[$mtype][] = $mime;
            }

            if ( count( $m ) > 1 ) {
                $main = $m[0];
                $mCount = count( $m );
                for ( $i = 1; $i < $mCount; $i += 1 ) {
                    $mime = $m[$i];
                    $this->mimeTypeAliases[$mime] = $main;
                }
            }
        }
    }

    public function setLogger( LoggerInterface $logger ) {
        $this->logger = $logger;
    }

    public function addExtraTypes( $types ) {
        $this->extraTypes .= "\n" . $types;
    }

    public function addExtraInfo( $info ) {
        $this->extraInfo .= "\n" . $info;
    }

    public function getExtensionsForType( $mime ) {
        $mime = strtolower( $mime );

        // Check the mime-to-ext map
        if ( isset( $this->mimetoExt[$mime] ) ) {
            return $this->mimetoExt[$mime];
        }

        // Resolve the MIME type to the canonical type
        if ( isset( $this->mimeTypeAliases[$mime] ) ) {
            $mime = $this->mimeTypeAliases[$mime];
            if ( isset( $this->mimetoExt[$mime] ) ) {
                return $this->mimetoExt[$mime];
            }
        }

        return null;
    }

    public function getTypesForExtension( $ext ) {
        $ext = strtolower( $ext );

        $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
        return $r;
    }

    public function guessTypesForExtension( $ext ) {
        $m = $this->getTypesForExtension( $ext );
        if ( is_null( $m ) ) {
            return null;
        }

        // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
        $m = trim( $m );
        $m = preg_replace( '/\s.*$/', '', $m );

        return $m;
    }

    public function isMatchingExtension( $extension, $mime ) {
        $ext = $this->getExtensionsForType( $mime );

        if ( !$ext ) {
            return null; // Unknown MIME type
        }

        $ext = explode( ' ', $ext );

        $extension = strtolower( $extension );
        return in_array( $extension, $ext );
    }

    public function isPHPImageType( $mime ) {
        // As defined by imagegetsize and image_type_to_mime
        static $types = [
            'image/gif', 'image/jpeg', 'image/png',
            'image/x-bmp', 'image/xbm', 'image/tiff',
            'image/jp2', 'image/jpeg2000', 'image/iff',
            'image/xbm', 'image/x-xbitmap',
            'image/vnd.wap.wbmp', 'image/vnd.xiff',
            'image/x-photoshop',
            'application/x-shockwave-flash',
        ];

        return in_array( $mime, $types );
    }

    function isRecognizableExtension( $extension ) {
        static $types = [
            // Types recognized by getimagesize()
            'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
            'bmp', 'tiff', 'tif', 'jpc', 'jp2',
            'jpx', 'jb2', 'swc', 'iff', 'wbmp',
            'xbm',

            // Formats we recognize magic numbers for
            'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
            'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
            'webp',

            // XML formats we sure hope we recognize reliably
            'svg',
        ];
        return in_array( strtolower( $extension ), $types );
    }

    public function improveTypeFromExtension( $mime, $ext ) {
        if ( $mime === 'unknown/unknown' ) {
            if ( $this->isRecognizableExtension( $ext ) ) {
                $this->logger->info( __METHOD__ . ': refusing to guess mime type for .' .
                    "$ext file, we should have recognized it\n" );
            } else {
                // Not something we can detect, so simply
                // trust the file extension
                $mime = $this->guessTypesForExtension( $ext );
            }
        } elseif ( $mime === 'application/x-opc+zip' ) {
            if ( $this->isMatchingExtension( $ext, $mime ) ) {
                // A known file extension for an OPC file,
                // find the proper MIME type for that file extension
                $mime = $this->guessTypesForExtension( $ext );
            } else {
                $this->logger->info( __METHOD__ .
                    ": refusing to guess better type for $mime file, " .
                    ".$ext is not a known OPC extension.\n" );
                $mime = 'application/zip';
            }
        } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
            // Textual types are sometimes not recognized properly.
            // If detected as text/plain, and has an extension which is textual
            // improve to the extension's type. For example, csv and json are often
            // misdetected as text/plain.
            $mime = $this->guessTypesForExtension( $ext );
        }

        # Media handling extensions can improve the MIME detected
        $callback = $this->extCallback;
        if ( $callback ) {
            $callback( $this, $ext, $mime /* by reference */ );
        }

        if ( isset( $this->mimeTypeAliases[$mime] ) ) {
            $mime = $this->mimeTypeAliases[$mime];
        }

        $this->logger->info( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
        return $mime;
    }

    public function guessMimeType( $file, $ext = true ) {
        if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
            $this->logger->info( __METHOD__ .
                ": WARNING: use of the \$ext parameter is deprecated. " .
                "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
        }

        $mime = $this->doGuessMimeType( $file, $ext );

        if ( !$mime ) {
            $this->logger->info( __METHOD__ .
                ": internal type detection failed for $file (.$ext)...\n" );
            $mime = $this->detectMimeType( $file, $ext );
        }

        if ( isset( $this->mimeTypeAliases[$mime] ) ) {
            $mime = $this->mimeTypeAliases[$mime];
        }

        $this->logger->info( __METHOD__ . ": guessed mime type of $file: $mime\n" );
        return $mime;
    }

    private function doGuessMimeType( $file, $ext ) {
        // Read a chunk of the file
        MediaWiki\suppressWarnings();
        $f = fopen( $file, 'rb' );
        MediaWiki\restoreWarnings();

        if ( !$f ) {
            return 'unknown/unknown';
        }

        $fsize = filesize( $file );
        if ( $fsize === false ) {
            return 'unknown/unknown';
        }

        $head = fread( $f, 1024 );
        $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR
        if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) {
            throw new UnexpectedValueException(
                "Seeking $tailLength bytes from EOF failed in " . __METHOD__ );
        }
        $tail = $tailLength ? fread( $f, $tailLength ) : '';
        fclose( $f );

        $this->logger->info( __METHOD__ .
            ": analyzing head and tail of $file for magic numbers.\n" );

        // Hardcode a few magic number checks...
        $headers = [
            // Multimedia...
            'MThd'             => 'audio/midi',
            'OggS'             => 'application/ogg',

            // Image formats...
            // Note that WMF may have a bare header, no magic number.
            "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
            "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
            '%PDF'             => 'application/pdf',
            'gimp xcf'         => 'image/x-xcf',

            // Some forbidden fruit...
            'MZ'               => 'application/octet-stream', // DOS/Windows executable
            "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
            "\x7fELF"          => 'application/octet-stream', // ELF binary
        ];

        foreach ( $headers as $magic => $candidate ) {
            if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
                $this->logger->info( __METHOD__ .
                    ": magic header in $file recognized as $candidate\n" );
                return $candidate;
            }
        }

        /* Look for WebM and Matroska files */
        if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
            $doctype = strpos( $head, "\x42\x82" );
            if ( $doctype ) {
                // Next byte is datasize, then data (sizes larger than 1 byte are stupid muxers)
                $data = substr( $head, $doctype + 3, 8 );
                if ( strncmp( $data, "matroska", 8 ) == 0 ) {
                    $this->logger->info( __METHOD__ . ": recognized file as video/x-matroska\n" );
                    return "video/x-matroska";
                } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
                    $this->logger->info( __METHOD__ . ": recognized file as video/webm\n" );
                    return "video/webm";
                }
            }
            $this->logger->info( __METHOD__ . ": unknown EBML file\n" );
            return "unknown/unknown";
        }

        /* Look for WebP */
        if ( strncmp( $head, "RIFF", 4 ) == 0 &&
            strncmp( substr( $head, 8, 7 ), "WEBPVP8", 7 ) == 0
        ) {
            $this->logger->info( __METHOD__ . ": recognized file as image/webp\n" );
            return "image/webp";
        }

        if ( ( strpos( $head, '<?php' ) !== false ) ||
            ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
            ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
            ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
            ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
            ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {

            $this->logger->info( __METHOD__ . ": recognized $file as application/x-php\n" );
            return 'application/x-php';
        }

        $xml = new XmlTypeCheck( $file );
        if ( $xml->wellFormed ) {
            $xmlTypes = $this->xmlTypes;
            if ( isset( $xmlTypes[$xml->getRootElement()] ) ) {
                return $xmlTypes[$xml->getRootElement()];
            } else {
                return 'application/xml';
            }
        }

        $script_type = null;

        # detect by shebang
        if ( substr( $head, 0, 2 ) == "#!" ) {
            $script_type = "ASCII";
        } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
            $script_type = "UTF-8";
        } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
            $script_type = "UTF-16BE";
        } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
            $script_type = "UTF-16LE";
        }

        if ( $script_type ) {
            if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
                // Quick and dirty fold down to ASCII!
                $pack = [ 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' ];
                $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
                $head = '';
                foreach ( $chars as $codepoint ) {
                    if ( $codepoint < 128 ) {
                        $head .= chr( $codepoint );
                    } else {
                        $head .= '?';
                    }
                }
            }

            $match = [];

            if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
                $mime = "application/x-{$match[2]}";
                $this->logger->info( __METHOD__ . ": shell script recognized as $mime\n" );
                return $mime;
            }
        }

        // Check for ZIP variants (before getimagesize)
        if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
            $this->logger->info( __METHOD__ . ": ZIP header present in $file\n" );
            return $this->detectZipType( $head, $tail, $ext );
        }

        MediaWiki\suppressWarnings();
        $gis = getimagesize( $file );
        MediaWiki\restoreWarnings();

        if ( $gis && isset( $gis['mime'] ) ) {
            $mime = $gis['mime'];
            $this->logger->info( __METHOD__ . ": getimagesize detected $file as $mime\n" );
            return $mime;
        }

        # Media handling extensions can guess the MIME by content
        # It's intentionally here so that if core is wrong about a type (false positive),
        # people will hopefully nag and submit patches :)
        $mime = false;
        # Some strings by reference for performance - assuming well-behaved hooks
        $callback = $this->guessCallback;
        if ( $callback ) {
            $callback( $this, $head, $tail, $file, $mime /* by reference */ );
        };

        return $mime;
    }

    function detectZipType( $header, $tail = null, $ext = false ) {
        if ( $ext ) { # TODO: remove $ext param
            $this->logger->info( __METHOD__ .
                ": WARNING: use of the \$ext parameter is deprecated. " .
                "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
        }

        $mime = 'application/zip';
        $opendocTypes = [
            'chart-template',
            'chart',
            'formula-template',
            'formula',
            'graphics-template',
            'graphics',
            'image-template',
            'image',
            'presentation-template',
            'presentation',
            'spreadsheet-template',
            'spreadsheet',
            'text-template',
            'text-master',
            'text-web',
            'text' ];

        // http://lists.oasis-open.org/archives/office/200505/msg00006.html
        $types = '(?:' . implode( '|', $opendocTypes ) . ')';
        $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";

        $openxmlRegex = "/^\[Content_Types\].xml/";

        if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
            $mime = $matches[1];
            $this->logger->info( __METHOD__ . ": detected $mime from ZIP archive\n" );
        } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
            $mime = "application/x-opc+zip";
            # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
            if ( $ext !== true && $ext !== false ) {
                if ( $this->isMatchingExtension( $ext, $mime ) ) {
                    /* A known file extension for an OPC file,
                     * find the proper mime type for that file extension
                     */
                    $mime = $this->guessTypesForExtension( $ext );
                } else {
                    $mime = "application/zip";
                }
            }
            $this->logger->info( __METHOD__ .
                ": detected an Open Packaging Conventions archive: $mime\n" );
        } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
            ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
            preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
            if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
                $mime = "application/msword";
            }
            switch ( substr( $header, 512, 6 ) ) {
                case "\xEC\xA5\xC1\x00\x0E\x00":
                case "\xEC\xA5\xC1\x00\x1C\x00":
                case "\xEC\xA5\xC1\x00\x43\x00":
                    $mime = "application/vnd.ms-powerpoint";
                    break;
                case "\xFD\xFF\xFF\xFF\x10\x00":
                case "\xFD\xFF\xFF\xFF\x1F\x00":
                case "\xFD\xFF\xFF\xFF\x22\x00":
                case "\xFD\xFF\xFF\xFF\x23\x00":
                case "\xFD\xFF\xFF\xFF\x28\x00":
                case "\xFD\xFF\xFF\xFF\x29\x00":
                case "\xFD\xFF\xFF\xFF\x10\x02":
                case "\xFD\xFF\xFF\xFF\x1F\x02":
                case "\xFD\xFF\xFF\xFF\x22\x02":
                case "\xFD\xFF\xFF\xFF\x23\x02":
                case "\xFD\xFF\xFF\xFF\x28\x02":
                case "\xFD\xFF\xFF\xFF\x29\x02":
                    $mime = "application/vnd.msexcel";
                    break;
            }

            $this->logger->info( __METHOD__ .
                ": detected a MS Office document with OPC trailer\n" );
        } else {
            $this->logger->info( __METHOD__ . ": unable to identify type of ZIP archive\n" );
        }
        return $mime;
    }

    private function detectMimeType( $file, $ext = true ) {
        if ( $ext ) {
            $this->logger->info( __METHOD__ .
                ": WARNING: use of the \$ext parameter is deprecated. "
                . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
        }

        $callback = $this->detectCallback;
        $m = null;
        if ( $callback ) {
            $m = $callback( $file );
        } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
            $mime_magic_resource = finfo_open( FILEINFO_MIME );

            if ( $mime_magic_resource ) {
                $m = finfo_file( $mime_magic_resource, $file );
                finfo_close( $mime_magic_resource );
            } else {
                $this->logger->info( __METHOD__ .
                    ": finfo_open failed on " . FILEINFO_MIME . "!\n" );
            }
        } else {
            $this->logger->info( __METHOD__ . ": no magic mime detector found!\n" );
        }

        if ( $m ) {
            # normalize
            $m = preg_replace( '![;, ].*$!', '', $m ); # strip charset, etc
            $m = trim( $m );
            $m = strtolower( $m );

            if ( strpos( $m, 'unknown' ) !== false ) {
                $m = null;
            } else {
                $this->logger->info( __METHOD__ . ": magic mime type of $file: $m\n" );
                return $m;
            }
        }

        // If desired, look at extension as a fallback.
        if ( $ext === true ) {
            $i = strrpos( $file, '.' );
            $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
        }
        if ( $ext ) {
            if ( $this->isRecognizableExtension( $ext ) ) {
                $this->logger->info( __METHOD__ . ": refusing to guess mime type for .$ext file, "
                    . "we should have recognized it\n" );
            } else {
                $m = $this->guessTypesForExtension( $ext );
                if ( $m ) {
                    $this->logger->info( __METHOD__ . ": extension mime type of $file: $m\n" );
                    return $m;
                }
            }
        }

        // Unknown type
        $this->logger->info( __METHOD__ . ": failed to guess mime type for $file!\n" );
        return 'unknown/unknown';
    }

    function getMediaType( $path = null, $mime = null ) {
        if ( !$mime && !$path ) {
            return MEDIATYPE_UNKNOWN;
        }

        // If MIME type is unknown, guess it
        if ( !$mime ) {
            $mime = $this->guessMimeType( $path, false );
        }

        // Special code for ogg - detect if it's video (theora),
        // else label it as sound.
        if ( $mime == 'application/ogg' && file_exists( $path ) ) {

            // Read a chunk of the file
            $f = fopen( $path, "rt" );
            if ( !$f ) {
                return MEDIATYPE_UNKNOWN;
            }
            $head = fread( $f, 256 );
            fclose( $f );

            $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) );

            // This is an UGLY HACK, file should be parsed correctly
            if ( strpos( $head, 'theora' ) !== false ) {
                return MEDIATYPE_VIDEO;
            } elseif ( strpos( $head, 'vorbis' ) !== false ) {
                return MEDIATYPE_AUDIO;
            } elseif ( strpos( $head, 'flac' ) !== false ) {
                return MEDIATYPE_AUDIO;
            } elseif ( strpos( $head, 'speex' ) !== false ) {
                return MEDIATYPE_AUDIO;
            } else {
                return MEDIATYPE_MULTIMEDIA;
            }
        }

        $type = null;
        // Check for entry for full MIME type
        if ( $mime ) {
            $type = $this->findMediaType( $mime );
            if ( $type !== MEDIATYPE_UNKNOWN ) {
                return $type;
            }
        }

        // Check for entry for file extension
        if ( $path ) {
            $i = strrpos( $path, '.' );
            $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );

            // TODO: look at multi-extension if this fails, parse from full path
            $type = $this->findMediaType( '.' . $e );
            if ( $type !== MEDIATYPE_UNKNOWN ) {
                return $type;
            }
        }

        // Check major MIME type
        if ( $mime ) {
            $i = strpos( $mime, '/' );
            if ( $i !== false ) {
                $major = substr( $mime, 0, $i );
                $type = $this->findMediaType( $major );
                if ( $type !== MEDIATYPE_UNKNOWN ) {
                    return $type;
                }
            }
        }

        if ( !$type ) {
            $type = MEDIATYPE_UNKNOWN;
        }

        return $type;
    }

    function findMediaType( $extMime ) {
        if ( strpos( $extMime, '.' ) === 0 ) {
            // If it's an extension, look up the MIME types
            $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
            if ( !$m ) {
                return MEDIATYPE_UNKNOWN;
            }

            $m = explode( ' ', $m );
        } else {
            // Normalize MIME type
            if ( isset( $this->mimeTypeAliases[$extMime] ) ) {
                $extMime = $this->mimeTypeAliases[$extMime];
            }

            $m = [ $extMime ];
        }

        foreach ( $m as $mime ) {
            foreach ( $this->mediaTypes as $type => $codes ) {
                if ( in_array( $mime, $codes, true ) ) {
                    return $type;
                }
            }
        }

        return MEDIATYPE_UNKNOWN;
    }

    public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
        $ca = $this->getIEContentAnalyzer();
        return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
    }

    protected function getIEContentAnalyzer() {
        if ( is_null( $this->IEAnalyzer ) ) {
            $this->IEAnalyzer = new IEContentAnalyzer;
        }
        return $this->IEAnalyzer;
    }
}