pdfbox application fat jar gives “Cannot read JBIG2 image: jbig2-imageio is not installed” but works ok running from IDE

Question

I am facing a problem when I build an application that uses pdfbox. The application is able to read books with jbig2 images when I run it from IDE (I use netbeans 8.1) (I have maven dependencies for jbig2 in pom.xml). The problem is when I build the application creating a fat jar. When I run the fat jar with the same input pdf, it gives the following error:

“Cannot read JBIG2 image: jbig2-imageio is not installed”

The threads that comment that error, do not seem to solve my problem (they say that a maven dependency has to be added to pom, but that dependency is already on my pom).

I have also checked that jbig2 library classes are inside the fat jar, so I have no idea of what is happening.

I have isolated the problem in a tinny application that looks like this:

public static void main( String[] args )
{
    String fileName = null;
    if( args.length == 0 )
    {
        fileName = "test.pdf";
    }
    else
    {
        fileName = args[0];
    }

    PdfDocumentWrapper doc = null;
    try
    {
        PdfboxFactory factory = new PdfboxFactory();
        doc = factory.createPdfDocumentWrapper();
        doc.loadPdf( fileName );
        for( int ii = 0; ii < doc.getNumberOfPages(); ii++ )
        {
            int pageNum = ii+1;
            System.out.println("\n\nProcessing page: " + pageNum +"\n---------------------------------");
            List<ImageWrapper> imageList = doc.getImagesOfPage(ii);

            int jj=0;
            for( ImageWrapper image: imageList )
            {
                jj++;
                System.out.println(String.format("  Page[%d]. Image[%d] -> bounds: %s",
                        pageNum, jj, image.getBounds().toString() ) );
            }
        }
    }
    catch( Exception ex )
    {
        ex.printStackTrace();
    }
    finally
    {
        if( doc != null )
        {
            try
            {
                doc.close();
            }
            catch( Exception ex )
            {
                ex.printStackTrace();
            }
        }
    }
}

I have placed the whole isolated example project here (with the purpose to help to solve the issue): http://www.frojasg1.com/20200504.PdfImageExtractor.zip

When I run that application from IDE, it produces the following output:

Processing page: 1
---------------------------------
  Page[1]. Image[1] -> bounds: java.awt.Rectangle[x=17,y=33,width=442,height=116]
  Page[1]. Image[2] -> bounds: java.awt.Rectangle[x=53,y=513,width=376,height=124]
  Page[1]. Image[3] -> bounds: java.awt.Rectangle[x=101,y=250,width=285,height=5]
------------------------------------------------------------------------

When I run the application from command line, it gives the following output:

$ java -jar ./PdfImageExtractor-v1.0-SNAPSHOT-all.jar


Processing page: 1
---------------------------------
may 04, 2020 3:40:18 PM org.apache.pdfbox.contentstream.PDFStreamEngine operatorException
GRAVE: Cannot read JBIG2 image: jbig2-imageio is not installed
may 04, 2020 3:40:18 PM org.apache.pdfbox.contentstream.PDFStreamEngine operatorException
GRAVE: Cannot read JBIG2 image: jbig2-imageio is not installed
may 04, 2020 3:40:18 PM org.apache.pdfbox.contentstream.PDFStreamEngine operatorException
GRAVE: Cannot read JBIG2 image: jbig2-imageio is not installed

Does anybody know why the fat jar is not able to read jbig2 images?

Answer 1

I posted the same question in pdfbox users mailing list, and here is the answer:

Your fat-jar consists of several ImageIO libs. You are simply merging all files 
to one big jar and overwriting the config files of those ImageIO libs. Have a 
look at the directory "/META-INF/services". The files of the JBig" plugin are 
overwritten by files of another plugin. Either you merge those files or don't 
create one big jar of all deps.

And the solution:

Thank you very much, it was that!

I have been able to create these META-INF files:

$ find src/serviceManifests/
src/serviceManifests/
src/serviceManifests/META-INF
src/serviceManifests/META-INF/services
src/serviceManifests/META-INF/services/javax.imageio.spi.ImageReaderSpi
src/serviceManifests/META-INF/services/javax.imageio.spi.ImageWriterSpi

merging them from the ones in ImageIO jars

By adding these lines to pom.xml:

<properties>
    <service.declaration.dir>src/serviceManifests</service.declaration.dir>
    <service.files.path>META-INF/services</service.files.path>
</properties>

<build>
    <resources>
    ...
        <resource>
            <directory>${service.declaration.dir}</directory>
            <includes>
                <include>${service.files.path}/*</include>
            </includes>
        </resource>
    ...
    </resources>
    ...
</build>

Problem solved.

Answer 2

add ImageIO.scanForPlugins(); before ImageIO.scanForPlugins(); can also solve this problem

Answer 3

If anyone stumbles here for the same error: Cannot read JBIG2 image: jbig2-imageio is not installed when reading a PDF file with Apache PDFBox, first you need to make sure you have an additional org.apache.pdfbox:jbig2-imageio dependency in your classpath.

Then you may also require calling ImageIO.scanForPlugins() as noted in the README.md of the pdfbox-jbig2 library (in my case it was not necessary).

For more information refer to the section on JAI Image I/O in PDFBox docs that covers reading JBIG2 images as well.