BitmapMetadataHandler.php

Go to the documentation of this file.

<?php
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use Wikimedia\RequestTimeout\TimeoutException;
use Wikimedia\XMPReader\Reader as XMPReader;
 
class BitmapMetadataHandler {
    private $metadata = [];
 
    private $metaPriority = [
        20 => [ 'other' ],
        40 => [ 'native' ],
        60 => [ 'iptc-good-hash', 'iptc-no-hash' ],
        70 => [ 'xmp-deprecated' ],
        80 => [ 'xmp-general' ],
        90 => [ 'xmp-exif' ],
        100 => [ 'iptc-bad-hash' ],
        120 => [ 'exif' ],
    ];
 
    private $iptcType = 'iptc-no-hash';
 
    private function doApp13( $app13 ) {
        try {
            $this->iptcType = JpegMetadataExtractor::doPSIR( $app13 );
        } catch ( TimeoutException $e ) {
            throw $e;
        } catch ( Exception $e ) {
            // Error reading the iptc hash information.
            // This probably means the App13 segment is something other than what we expect.
            // However, still try to read it, and treat it as if the hash didn't exist.
            wfDebug( "Error parsing iptc data of file: " . $e->getMessage() );
            $this->iptcType = 'iptc-no-hash';
        }
 
        $iptc = IPTC::parse( $app13 );
        $this->addMetadata( $iptc, $this->iptcType );
    }
 
    public function getExif( $filename, $byteOrder ) {
        $showEXIF = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::ShowEXIF );
        if ( file_exists( $filename ) && $showEXIF ) {
            $exif = new Exif( $filename, $byteOrder );
            $data = $exif->getFilteredData();
            if ( $data ) {
                $this->addMetadata( $data, 'exif' );
            }
        }
    }
 
    public function addMetadata( $metaArray, $type = 'other' ) {
        if ( isset( $this->metadata[$type] ) ) {
            /* merge with old data */
            $metaArray += $this->metadata[$type];
        }
 
        $this->metadata[$type] = $metaArray;
    }
 
    public function getMetadataArray() {
        // this seems a bit ugly... This is all so its merged in right order
        // based on the MWG recommendation.
        $temp = [];
        krsort( $this->metaPriority );
        foreach ( $this->metaPriority as $pri ) {
            foreach ( $pri as $type ) {
                if ( isset( $this->metadata[$type] ) ) {
                    // Do some special casing for multilingual values.
                    // Don't discard translations if also as a simple value.
                    foreach ( $this->metadata[$type] as $itemName => $item ) {
                        if ( is_array( $item ) && isset( $item['_type'] ) && $item['_type'] === 'lang' &&
                            isset( $temp[$itemName] ) && !is_array( $temp[$itemName] )
                        ) {
                            $default = $temp[$itemName];
                            $temp[$itemName] = $item;
                            $temp[$itemName]['x-default'] = $default;
                            unset( $this->metadata[$type][$itemName] );
                        }
                    }
 
                    $temp += $this->metadata[$type];
                }
            }
        }
 
        return $temp;
    }
 
    public static function Jpeg( $filename ) {
        $showXMP = XMPReader::isSupported();
        $meta = new self();
 
        $seg = JpegMetadataExtractor::segmentSplitter( $filename );
 
        if ( isset( $seg['SOF'] ) ) {
            $meta->addMetadata( [ 'SOF' => $seg['SOF'] ] );
        }
        if ( isset( $seg['COM'] ) && isset( $seg['COM'][0] ) ) {
            $meta->addMetadata( [ 'JPEGFileComment' => $seg['COM'] ], 'native' );
        }
        if ( isset( $seg['PSIR'] ) && count( $seg['PSIR'] ) > 0 ) {
            foreach ( $seg['PSIR'] as $curPSIRValue ) {
                $meta->doApp13( $curPSIRValue );
            }
        }
        if ( isset( $seg['XMP'] ) && $showXMP ) {
            $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
            $xmp->parse( $seg['XMP'] );
            foreach ( $seg['XMP_ext'] as $xmpExt ) {
                /* Support for extended xmp in jpeg files
                 * is not well tested and a bit fragile.
                 */
                $xmp->parseExtended( $xmpExt );
            }
            $res = $xmp->getResults();
            foreach ( $res as $type => $array ) {
                $meta->addMetadata( $array, $type );
            }
        }
 
        $meta->getExif( $filename, $seg['byteOrder'] ?? 'BE' );
 
        return $meta->getMetadataArray();
    }
 
    public static function PNG( $filename ) {
        $showXMP = XMPReader::isSupported();
 
        $meta = new self();
        $array = PNGMetadataExtractor::getMetadata( $filename );
        if ( isset( $array['text']['xmp']['x-default'] )
            && $array['text']['xmp']['x-default'] !== '' && $showXMP
        ) {
            $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
            $xmp->parse( $array['text']['xmp']['x-default'] );
            $xmpRes = $xmp->getResults();
            foreach ( $xmpRes as $type => $xmpSection ) {
                $meta->addMetadata( $xmpSection, $type );
            }
        }
        unset( $array['text']['xmp'] );
        // @phan-suppress-next-line PhanTypePossiblyInvalidDimOffset xmp is not alone in text, makes text always set
        $meta->addMetadata( $array['text'], 'native' );
        unset( $array['text'] );
        $array['metadata'] = $meta->getMetadataArray();
        $array['metadata']['_MW_PNG_VERSION'] = PNGMetadataExtractor::VERSION;
 
        return $array;
    }
 
    public static function GIF( $filename ) {
        $meta = new self();
        $baseArray = GIFMetadataExtractor::getMetadata( $filename );
 
        if ( count( $baseArray['comment'] ) > 0 ) {
            $meta->addMetadata( [ 'GIFFileComment' => $baseArray['comment'] ], 'native' );
        }
 
        if ( $baseArray['xmp'] !== '' && XMPReader::isSupported() ) {
            $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
            $xmp->parse( $baseArray['xmp'] );
            $xmpRes = $xmp->getResults();
            foreach ( $xmpRes as $type => $xmpSection ) {
                $meta->addMetadata( $xmpSection, $type );
            }
        }
 
        unset( $baseArray['comment'] );
        unset( $baseArray['xmp'] );
 
        $baseArray['metadata'] = $meta->getMetadataArray();
        $baseArray['metadata']['_MW_GIF_VERSION'] = GIFMetadataExtractor::VERSION;
 
        return $baseArray;
    }
 
    public static function Tiff( $filename ) {
        if ( file_exists( $filename ) ) {
            $byteOrder = self::getTiffByteOrder( $filename );
            if ( !$byteOrder ) {
                throw new MWException( "Error determining byte order of $filename" );
            }
            $exif = new Exif( $filename, $byteOrder );
            $data = $exif->getFilteredData();
            if ( $data ) {
                $data['MEDIAWIKI_EXIF_VERSION'] = Exif::version();
 
                return $data;
            } else {
                throw new MWException( "Could not extract data from tiff file $filename" );
            }
        } else {
            throw new MWException( "File doesn't exist - $filename" );
        }
    }
 
    public static function getTiffByteOrder( $filename ) {
        $fh = fopen( $filename, 'rb' );
        if ( !$fh ) {
            return false;
        }
        $head = fread( $fh, 2 );
        fclose( $fh );
 
        switch ( $head ) {
            case 'II':
                return 'LE'; // II for intel.
            case 'MM':
                return 'BE'; // MM for motorla.
            default:
                return false; // Something went wrong.
 
        }
    }
}