Catch output stream of xsl result-document

Question

I need a way to interfere in writting xsl result documents to avoid writting them to file system. Right now my template is writting to a temporary directory, and then i zip that directory. I want to do that whitout writting to file system. I am using saxon procesor. A solution that is using just standard java libraries is perfered. Any suggestion is appreciated.

EDIT: I found this class for .net saxon api http://www.saxonica.com/documentation/dotnetdoc/Saxon/Api/IResultDocumentHandler.html I need something equivalent for java.

Answer 1

You need to implement interface net.sf.saxon.OutputURIResolver
http://www.saxonica.com/documentation/javadoc/net/sf/saxon/lib/OutputURIResolver.html
You can redirect output in resolve method however you like. In my case this is how implemented class looks like.

public class ZipOutputURIReslover implements OutputURIResolver{

    private ZipOutputStream zipOut;

    public ZipOutputURIReslover(ZipOutputStream zipOut) {
        super();
        this.zipOut = zipOut;
    }

    public void close(Result arg0) throws TransformerException {
        try {
            zipOut.closeEntry();
        } catch (IOException e) {
            e.printStackTrace();
        }
    }

    public Result resolve(String href, String base) throws TransformerException {
        try {
            zipOut.putNextEntry(new ZipEntry(href));
        } catch (IOException e) {
            e.printStackTrace();
        }
        return new StreamResult(zipOut);
    }
}

After this you need to register net.sf.saxon.OutputURIResolver to trasnformer factory.

ZipOutputStream zipOut = new ZipOutputStream(new FileOutputStream("file.zip"));
factory.setAttribute("http://saxon.sf.net/feature/outputURIResolver", new ZipOutputURIReslover(zipOut));

When you load your template and run transformation all xsl:result-documents will be written directly to zipOutputStream.

Answer was found here http://sourceforge.net/p/saxon/discussion/94027/thread/9ee79dea/#70a9/6fef

Answer 2

You can use new StreamResult(ByteArrayOutputStream()) to catch xslt output, if you don't want to write to files, and then you can save memory data from byte array to zip file using this approach In Java: How to zip file from byte[] array?

Answer 3

Note that recent versions of Saxon require that the href (URI) is unique. Thus the system ID of the stream in the output resolver must also be unique.

For example:

1. Specify the result document href values in the stylesheet

   <xsl:result-document href="{$filename}" method="text">
   

2. Create the output resolver

   public class ZipOutputURIReslover implements OutputURIResolver{
   
       private ZipOutputStream zipOut;
   
       public ZipOutputURIReslover(ZipOutputStream zipOut) {
           super();
           this.zipOut = zipOut;
       }
   
       public void close(Result arg0) throws TransformerException {
           try {
               zipOut.closeEntry();
           } catch (IOException e) {
               e.printStackTrace();
           }
       }
   
       public Result resolve(String href, String base) throws TransformerException {
           try {
               zipOut.putNextEntry(new ZipEntry(href));
           } catch (IOException e) {
               e.printStackTrace();
           }
           Result result = new StreamResult(zipOut);
           // Must ensure the stream is given a unique ID
           result.setSystemId(UUID.randomUUID().toString());
           return result;
       }
   }
   

3. Attach the output resolver to the transformer

   ZipOutputURIResolver outputResolver = new ZipOutputURIResolver(outputStream);
   // Controller is the Saxon implementation of the JAXP Transformer
   ((Controller) transformer).setOutputURIResolver(outputResolver);