DSpace can apply filters or transformations to files/bitstreams, creating new content. Filters are included that extract text for full-text searching, and create thumbnails for items that contain images. The media filters are controlled by the
dspace filter-media script which traverses the asset store, invoking all configured
FormatFilter classes on files/bitstreams (see Configuring Media Filters for more information on how they are configured).
Below is a listing of all currently available Media Filters, and what they actually do:
Enabled by Default?
HTML Text Extractor
extracts the full text of HTML documents for full text indexing. (Uses Swing's HTML Parser)
creates thumbnail images of GIF, JPEG and PNG files
Branded Preview JPEG
creates a branded preview image for GIF, JPEG and PNG files
PDF Text Extractor
extracts the full text of Adobe PDF documents (only if text-based or OCRed) for full text indexing. (Uses the Apache PDFBox tool)
XPDF Text Extractor
extracts the full text of Adobe PDF documents (only if text-based or OCRed) for full text indexing (Uses the XPDF command line tools available for Unix.) See XPDF Filter Configuration for details on installing/enabling.
Word Text Extractor
extracts the full text of Microsoft Word or Plain Text documents for full text indexing. (Uses the "Microsoft Word Text Mining" tools.)
PowerPoint Text Extractor
extracts the full text of slides and notes in Microsoft PowerPoint and PowerPoint XML documents for full text indexing (Uses the Apache POI tools.)
|ImageMagick Image Thumbnail Generator|
|uses ImageMagick to generate thumbnails for image bitstreams. Requires installation of ImageMagick on your server. See ImageMagick Media Filters.||false|
|ImageMagick PDF Thumbnail Generator||org.dspace.app.mediafilter.ImageMagickPdfThumbnailFilter||uses ImageMagick and Ghostscript to generate thumbnails for PDF bitstreams. Requires installation of ImageMagick and Ghostscript on your server. See ImageMagick Media Filters.||false|
Please note that the
filter-media script will automatically update the DSpace search index by default (see Legacy methods for re-indexing content) This is the recommended way to run these scripts. But, should you wish to disable it, you can pass the -n flag to either script to do so (see Executing (via Command Line) below).
The media filter plugin configuration
dspace.cfg contains a list of all enabled media/format filter plugins (see Configuring Media Filters for more information). By modifying the value of
filter.plugins you can disable or enable MediaFilter plugins.
The media filter system is intended to be run from the command line (or regularly as a cron task):
With no options, this traverses the asset store, applying media filters to bitstreams, and skipping bitstreams that have already been filtered.
Available Command-Line Options:
[dspace]/bin/dspace filter-media -h
[dspace]/bin/dspace filter-media -f
[dspace]/bin/dspace filter-media -i 123456789/2
[dspace]/bin/dspace filter-media -m 1000
[dspace]/bin/dspace filter-media -n
[dspace]/bin/dspace filter-media -p "PDF Text Extractor","Word Text Extractor"
[dspace]/bin/dspace filter-media -s 123456789/9,123456789/100
[dspace]/bin/dspace filter-media -s `less filter-skiplist.txt`
[dspace]/bin/dspace filter-media -v
org.dspace.app.mediafilter.FormatFilterinterface. See the Creating a new Media/Format Filter topic and comments in the source file
FormatFilter.javafor more information. In theory filters could be implemented in any programming language (C, Perl, etc.) However, they need to be invoked by the Java code in the Media Filter class that you create.
New Media Filters must implement the org.dspace.app.mediafilter.FormatFilter interface. More information on the methods you need to implement is provided in the FormatFilter.java source file. For example:
public class MySimpleMediaFilter implements FormatFilter
Alternatively, you could extend the org.dspace.app.mediafilter.MediaFilter class, which just defaults to performing no pre/post-processing of bitstreams before or after filtering.
public class MySimpleMediaFilter extends MediaFilter
You must give your new filter a "name", by adding it and its name to the plugin.named.org.dspace.app.mediafilter.FormatFilter field in dspace.cfg. In addition to naming your filter, make sure to specify its input formats in the filter.<class path>.inputFormats config item. Note the input formats must match the short description field in the Bitstream Format Registry (i.e. bitstreamformatregistry table).
plugin.named.org.dspace.app.mediafilter.FormatFilter = \ org.dspace.app.mediafilter.MySimpleMediaFilter = My Simple Text Filter, \ ... filter.org.dspace.app.mediafilter.MySimpleMediaFilter.inputFormats = Text
If you neglect to define the inputFormats for a particular filter, the MediaFilterManager will never call that filter, since it will never find a bitstream which has a format matching that filter's input format(s).
If you have a complex Media Filter class, which actually performs different filtering for different formats (e.g. conversion from Word to PDF and conversion from Excel to CSV), you should define this as described in Chapter 184.108.40.206 .
If you have a more complex Media/Format Filter, which actually performs multiple filtering or conversions for different formats (e.g. conversion from Word to PDF and conversion from Excel to CSV), you should have define a class which implements the FormatFilter interface, while also extending the Chapter 220.127.116.11 SelfNamedPlugin class. For example:
public class MyComplexMediaFilter extends SelfNamedPlugin implements FormatFilter
Since SelfNamedPlugins are self-named (as stated), they must provide the various names the plugin uses by defining a getPluginNames() method. Generally speaking, each "name" the plugin uses should correspond to a different type of filter it implements (e.g. "Word2PDF" and "Excel2CSV" are two good names for a complex media filter which performs both Word to PDF and Excel to CSV conversions).
Self-Named Media/Format Filters are also configured differently in dspace.cfg. Below is a general template for a Self Named Filter (defined by an imaginary MyComplexMediaFilter class, which can perform both Word to PDF and Excel to CSV conversions):
#Add to a list of all Self Named filters plugin.selfnamed.org.dspace.app.mediafilter.FormatFilter = \ org.dspace.app.mediafilter.MyComplexMediaFilter #Define input formats for each "named" plugin this filter implements filter.org.dspace.app.mediafilter.MyComplexMediaFilter.Word2PDF.inputFormats = Microsoft Word filter.org.dspace.app.mediafilter.MyComplexMediaFilter.Excel2CSV.inputFormats = Microsoft Excel
As shown above, each Self-Named Filter class must be listed in the
plugin.selfnamed.org.dspace.app.mediafilter.FormatFilter item in
dspace.cfg. In addition, each Self-Named Filter must define the input formats for each named plugin defined by that filter. In the above example the MyComplexMediaFilter class is assumed to have defined two named plugins,
Excel2CSV. So, these two valid plugin names ("Word2PDF" and "Excel2CSV") must be returned by the
getPluginNames() method of the
These named plugins take different input formats as defined above (see the corresponding inputFormats setting).
If you neglect to define the
For a particular Self-Named Filter, you are also welcome to define additional configuration settings in dspace.cfg. To continue with our current example, each of our imaginary plugins actually results in a different output format (Word2PDF creates "Adobe PDF", while Excel2CSV creates "Comma Separated Values"). To allow this complex Media Filter to be even more configurable (especially across institutions, with potential different "Bitstream Format Registries"), you may wish to allow for the output format to be customizable for each named plugin. For example:
#Define output formats for each named plugin filter.org.dspace.app.mediafilter.MyComplexMediaFilter.Word2PDF.output Format = Adobe PDF filter.org.dspace.app.mediafilter.MyComplexMediaFilter.Excel2CSV.outputFormat = Comma Separated Values
Any custom configuration fields in dspace.cfg defined by your filter are ignored by the MediaFilterManager, so it is up to your custom media filter class to read those configurations and apply them as necessary. For example, you could use the following sample Java code in your MyComplexMediaFilter class to read these custom outputFormat configurations from dspace.cfg:
#Get "outputFormat" configuration from dspace.cfg String outputFormat = ConfigurationManager.getProperty(MediaFilterManager.FILTER_PREFIX + "." + MyComplexMediaFilter.class.getName() + "." + this.getPluginInstanceName() + ".outputFormat");
|Example Value||filter.org.dspace.app.mediafilter.publicPermission = JPEGFilter, XPDF2Thumbnail|
|Informational Note||By default mediafilter derivatives / thumbnails inherit the same permissions of the parent bitstream, but you can override this, in case you want to make publicly accessible derivative / thumbnail content, typically the thumbnails of objects for the browse list. List the MediaFilter name's that would get public accessible permissions. Any media filters not listed will instead inherit the permissions of the parent bitstream.|