DjVuHandler.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Media;
 
use MediaWiki\FileRepo\File\File;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\PoolCounter\PoolCounterWorkViaCallback;
use MediaWiki\Shell\Shell;
 
class DjVuHandler extends ImageHandler {
    private const EXPENSIVE_SIZE_LIMIT = 10_485_760; // 10MiB
 
    // Constants for getHandlerState
    private const STATE_DJVU_IMAGE = 'djvuImage';
    private const STATE_TEXT_TREE = 'djvuTextTree';
    private const STATE_META_TREE = 'djvuMetaTree';
    private const CACHE_VERSION = 'v2';
 
    public function isEnabled() {
        $djvuRenderer = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuRenderer );
        $djvuDump = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuDump );
        if ( !$djvuRenderer || !$djvuDump ) {
            // @codeCoverageIgnoreStart
            wfDebug( "DjVu is disabled, please set \$wgDjvuRenderer and \$wgDjvuDump" );
 
            return false;
            // @codeCoverageIgnoreEnd
        }
        return true;
    }
 
    public function mustRender( $file ) {
        return true;
    }
 
    public function isExpensiveToThumbnail( $file ) {
        return $file->getSize() > static::EXPENSIVE_SIZE_LIMIT;
    }
 
    public function isMultiPage( $file ) {
        return true;
    }
 
    public function getParamMap() {
        return [
            'img_width' => 'width',
            'img_page' => 'page',
        ];
    }
 
    public function validateParam( $name, $value ) {
        if ( $name === 'page' ) {
            return is_int( $value ) ||
                // Extra junk on the end of page, probably actually a caption
                // e.g. [[File:Foo.djvu|thumb|Page 3 of the document shows foo]]
                ( is_string( $value ) && trim( $value ) === (string)intval( $value ) );
        }
        return in_array( $name, [ 'width', 'height', 'page', 'physicalWidth', 'physicalHeight' ] ) &&
            (int)$value > 0;
    }
 
    public function makeParamString( $params ) {
        $page = $params['page'] ?? 1;
        $width = $params['physicalWidth'] ?? $params['width'] ?? null;
        if ( !$width ) {
            return false;
        }
 
        return "page{$page}-{$width}px";
    }
 
    public function parseParamString( $str ) {
        $m = false;
        if ( preg_match( '/^page(\d+)-(\d+)px$/', $str, $m ) ) {
            return [ 'width' => $m[2], 'page' => $m[1] ];
        }
        return false;
    }
 
    public function normaliseParams( $image, &$params ) {
        // Copy-paste from TransformationalImageHandler::normaliseParams() and PdfHandler
        // Probably should be trait or subclass.
        if ( !parent::normaliseParams( $image, $params ) ) {
            return false;
        }
        $srcWidth = $image->getWidth( $params['page'] );
        $srcHeight = $image->getHeight( $params['page'] );
        $params['physicalWidth'] = $this->getSteppedThumbWidth(
            $image, $params['physicalWidth'], $srcWidth, $srcHeight
        );
        $params['physicalHeight'] = File::scaleHeight( $srcWidth, $srcHeight, $params['physicalWidth'] );
        return true;
    }
 
    protected function getScriptParams( $params ) {
        return [
            'width' => $params['width'],
            'page' => $params['page'],
        ];
    }
 
    public function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
        $djvuRenderer = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuRenderer );
        $djvuPostProcessor = MediaWikiServices::getInstance()->getMainConfig()
            ->get( MainConfigNames::DjvuPostProcessor );
        if ( !$this->normaliseParams( $image, $params ) ) {
            return new TransformParameterError( $params );
        }
        $width = $params['width'];
        $height = $params['height'];
        $page = $params['page'];
 
        if ( $flags & self::TRANSFORM_LATER ) {
            $params = [
                'width' => $width,
                'height' => $height,
                'page' => $page
            ];
 
            return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
        }
 
        if ( !wfMkdirParents( dirname( $dstPath ), null, __METHOD__ ) ) {
            // @codeCoverageIgnoreStart
            return new MediaTransformError(
                'thumbnail_error',
                $width,
                $height,
                wfMessage( 'thumbnail_dest_directory' )
            );
            // @codeCoverageIgnoreEnd
        }
 
        // Get local copy source for shell scripts
        // Thumbnail extraction is very inefficient for large files.
        // Provide a way to pool count limit the number of downloaders.
        if ( $image->getSize() >= 1e7 ) { // 10 MB
            $work = new PoolCounterWorkViaCallback( 'GetLocalFileCopy', sha1( $image->getName() ),
                [
                    'doWork' => static function () use ( $image ) {
                        return $image->getLocalRefPath();
                    }
                ]
            );
            $srcPath = $work->execute();
        } else {
            $srcPath = $image->getLocalRefPath();
        }
 
        if ( $srcPath === false ) {
            // Failed to get local copy
            // @codeCoverageIgnoreStart
            wfDebugLog( 'thumbnail',
                sprintf( 'Thumbnail failed on %s: could not get local copy of "%s"',
                    wfHostname(), $image->getName() ) );
 
            return new MediaTransformError( 'thumbnail_error',
                $params['width'], $params['height'],
                wfMessage( 'filemissing' )
            );
            // @codeCoverageIgnoreEnd
        }
 
        // Use a subshell (brackets) to aggregate stderr from both pipeline commands
        // before redirecting it to the overall stdout. This works in both Linux and Windows XP.
        $cmd = '(' . Shell::escape(
            $djvuRenderer,
            "-format=ppm",
            "-page={$page}",
            "-size={$params['physicalWidth']}x{$params['physicalHeight']}",
            $srcPath );
        if ( $djvuPostProcessor ) {
            $cmd .= " | {$djvuPostProcessor}";
        }
        $cmd .= ' > ' . Shell::escape( $dstPath ) . ') 2>&1';
        wfDebug( __METHOD__ . ": $cmd" );
        $shell = Shell::command()->unsafeCommand( $cmd )->execute();
        $retval = $shell->getExitCode();
        $err = $shell->getStderr();
 
        $removed = $this->removeBadFile( $dstPath, $retval );
        if ( ( $retval !== 0 || $removed ) && $retval !== null ) {
            // @codeCoverageIgnoreStart
            $this->logErrorForExternalProcess( $retval, $err, $cmd );
            return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
            // @codeCoverageIgnoreEnd
        }
        $params = [
            'width' => $width,
            'height' => $height,
            'page' => $page
        ];
 
        return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
    }
 
    private function getDjVuImage( MediaHandlerState $state, string $path ): DjVuImage {
        $deja = $state->getHandlerState( self::STATE_DJVU_IMAGE );
        if ( !$deja ) {
            $deja = new DjVuImage( $path );
            $state->setHandlerState( self::STATE_DJVU_IMAGE, $deja );
        }
        return $deja;
    }
 
    private function getMetadataInternal( File $file, bool $gettext ) {
        $itemNames = [ 'error', '_error', 'data' ];
        if ( $gettext ) {
            $itemNames[] = 'text';
        }
        $unser = $file->getMetadataItems( $itemNames );
 
        if ( isset( $unser['error'] ) ) {
            return false;
        }
        if ( isset( $unser['_error'] ) ) {
            return false;
        }
        return $unser;
    }
 
    public function getMetaTree( $image, $gettext = false ) {
        if ( $gettext && $image->getHandlerState( self::STATE_TEXT_TREE ) ) {
            return $image->getHandlerState( self::STATE_TEXT_TREE );
        }
        if ( !$gettext && $image->getHandlerState( self::STATE_META_TREE ) ) {
            return $image->getHandlerState( self::STATE_META_TREE );
        }
 
        $metadata = $this->getMetadataInternal( $image, $gettext );
        if ( !$metadata ) {
            return false;
        }
 
        if ( !$gettext ) {
            unset( $metadata['text'] );
        }
        return $metadata;
    }
 
    public function getThumbType( $ext, $mime, $params = null ) {
        $djvuOutputExtension = MediaWikiServices::getInstance()->getMainConfig()
            ->get( MainConfigNames::DjvuOutputExtension );
        static $djvuMime = null;
        if ( $djvuMime === null ) {
            $magic = MediaWikiServices::getInstance()->getMimeAnalyzer();
            $djvuMime = $magic->getMimeTypeFromExtensionOrNull( $djvuOutputExtension );
        }
 
        return [ $djvuOutputExtension, $djvuMime ];
    }
 
    public function getSizeAndMetadata( $state, $path ) {
        wfDebug( "Getting DjVu metadata for $path" );
 
        $djvuImage = $this->getDjVuImage( $state, $path );
        $metadata = $djvuImage->retrieveMetaData();
        if ( $metadata === false ) {
            // Special value so that we don't repetitively try and decode a broken file.
            $metadata = [ 'error' => 'Error extracting metadata' ];
        }
        return [ 'metadata' => $metadata ] + $djvuImage->getImageSize();
    }
 
    public function getMetadataType( $image ) {
        // historical reasons
        return 'djvuxml';
    }
 
    public function isFileMetadataValid( $image ) {
        return $image->getMetadataArray() ? self::METADATA_GOOD : self::METADATA_BAD;
    }
 
    public function pageCount( File $image ) {
        $info = $this->getDimensionInfo( $image );
 
        return $info ? $info['pageCount'] : false;
    }
 
    public function getPageDimensions( File $image, $page ) {
        $index = $page - 1; // MW starts pages at 1
 
        $info = $this->getDimensionInfo( $image );
        if ( $info && isset( $info['dimensionsByPage'][$index] ) ) {
            return $info['dimensionsByPage'][$index];
        }
 
        return false;
    }
 
    protected function getDimensionInfo( File $file ) {
        $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
        return $cache->getWithSetCallback(
            $cache->makeKey( 'file-djvu', 'dimensions', self::CACHE_VERSION, $file->getSha1() ),
            $cache::TTL_INDEFINITE,
            function () use ( $file ) {
                $tree = $this->getMetaTree( $file );
                return $this->getDimensionInfoFromMetaTree( $tree );
            },
            [ 'pcTTL' => $cache::TTL_INDEFINITE ]
        );
    }
 
    protected function getDimensionInfoFromMetaTree( $metatree ) {
        if ( !$metatree ) {
            return false;
        }
        $dimsByPage = [];
 
        if ( !isset( $metatree['data'] ) || !$metatree['data'] ) {
            return false;
        }
        foreach ( $metatree['data']['pages'] as $page ) {
            if ( !$page ) {
                $dimsByPage[] = false;
            } else {
                $dimsByPage[] = [
                    'width' => (int)$page['width'],
                    'height' => (int)$page['height'],
                ];
            }
        }
        return [
            'pageCount' => count( $metatree['data']['pages'] ),
            'dimensionsByPage' => $dimsByPage
        ];
    }
 
    public function getPageText( File $image, $page ) {
        $tree = $this->getMetaTree( $image, true );
        if ( !$tree ) {
            return false;
        }
        if ( isset( $tree['text'] ) && isset( $tree['text'][$page - 1] ) ) {
            return $tree['text'][$page - 1];
        }
        return false;
    }
 
    public function useSplitMetadata() {
        return true;
    }
}
 
class_alias( DjVuHandler::class, 'DjVuHandler' );