XSLT Streaming Chained Transform

Question

I am using Saxon EE to transform a very large document, using streaming transform. I now need to chain multiple XSLTs. How does one go about doing that ? When not streaming, I have used the XSLTTransformer class as destination, to do chained transforms. If I am not mistaken, I guess I cannot do that, as that would create a result tree as against result stream. Thanks, Ani

Answer 1

Pipe the SAX output events of the 1st transform into the SAX input events of the 2nd transform.

I've attached some example Scala code which shows this.

Basically it kicks off the 2nd XSLT first, which behind the scenes invokes the 1st XSLT with the initial input document capturing the intermediate output which is then fed as the input to the 2nd XSLT in real time. Nice.

• It uses JAXP interfaces, so no S9 API.
• I've successfully tested it with a 1.2GB input XML file.

Hope this helps.

XSLT 3.0 rocks by the way! Good choice.

import javax.xml.transform.sax.{SAXResult, SAXSource}
import javax.xml.transform.stream.{StreamResult, StreamSource}
import javax.xml.transform.{Source, Transformer}

import com.saxonica.config.StreamingTransformerFactory
import org.xml.sax._
import org.xml.sax.helpers.XMLFilterImpl

object Main extends App
{
  val transformer1 = transformer("transform-1.xsl")
  val transformer2 = transformer("transform-2.xsl")

  val inputXML = "big.xml"

  transformer2.transform(
    new SAXSource(
      new Transformer1OutputReader(transformer1, new StreamSource(inputXML)),
      null
    ),
    new StreamResult("out.xml")
  )

  def transformer(xslt : String) =
    new StreamingTransformerFactory().newTransformer(new StreamSource(xslt))

}


class Transformer1OutputReader(
  transformer1 : Transformer,
  source1 : Source) extends XMLFilterImpl
{
  def parseImpl() =
  {
    println("parseImpl()")

    val inputToSecondXslt : ContentHandler = getContentHandler

    transformer1.transform(
      source1,
      new SAXResult(inputToSecondXslt)
    )
  }

  override def parse(input : InputSource) = parseImpl
  override def parse(systemId : String) = parseImpl
  override def setFeature(name: String, value: Boolean) : Unit = {}
}