Framework User Guide

Relevant for Java developers. For using metafacture without Java Code see the FLUX user guide.

This page explains how to create a Metafacture objects and how to assemble them to form a processing pipeline. We use as an example a simple pipeline containing a Metafix instance.

Building a Flow

A Flow consists of a data source, an arbitrary number of pipe elements and finally a data sink. The individual elements are connected by calling the setReceiver() method. The following code snipped shows an example.

// create necessary objects
final PicaReader reader = new PicaReader();
final Metafix fix = new Metafix("definition.fix");
final ListMapWriter writer = new ListMapWriter();

//connect them

//start processing;

Note that the call setReceiver() returns its argument, preserving the respective type. Thus the calls can be chained to build up a pipeline as shown in the listing. Finally the processing is started by calling the respective method on the data source/reader. The method name depends on the reader. In the Metafacture project read() is used by convention.

The following code snippet shows a few more sophisticated connection patterns, such as adding an additional element, junctions or splitters.

//adding logging
reader.setReceiver(new LogPipe()).setReceiver(metafix).setReceiver(writer);

//adding a tee junction
reader.setReceiver(new Tee()).setReceivers(writer1, writer2);

// create e.g. three threads
reader.setReceiver(new ObjectThreader<>()).addReceiver(...).addReceiver(...).addReceiver(...);
//splitting based on a metamorph description
final Splitter splitter = new Splitter("morph/typeSplitter.xml");
reader.setReceiver(splitter).setReceiver("Tn", writer1);
splitter.setReceiver("Tp", writer2);

Piping different Objects

Objects as Eventstream

public interface StreamSender {
	<R extends StreamReceiver> R setReceiver(R streamReceiver);
public interface StreamReceiver {
	void startRecord(String identifier);
	void endRecord();
	void startEntity(String name);
	void endEntity();
	void literal(String name, String value);

Error Handling

If an exception occurs during the processing of a stream of records, it is back propagated to the first element in the chain. This normally means that processing is terminated which may not be the preferred action. Imagine processing a million records. One normally prefers to log any error but continue the processing. For this reason an error handler may be registered with the Metamorph object. It catches all exceptions occurring in the Metamorph object and below.

metamorph.setErrorHandler(new MetamorphErrorHandler() {
	public void error(final Exception e) {
		// TODO fill in your error handling code