Lunatech

Processing EDIFACT messages with Smooks

By Ludovico Fischer on 30 December 2011

Smooks is a set of tools written in the Java programming language to assist you in parsing a variety of formats such as XML, CSV, fixed width formats and making the information accessible from your Java code. My interest in Smooks was related to the fact that it is one of the few open source projects which offer facilities for accessing data in EDI formats. In fact, Smooks might be the only actively-maintained open source project that offers this.

In this article, I am going to share the results of my investigation on using Smooks to process EDI data, and in particular messages that follow the UN/EDIFACT standards.

What is UN/EDIFACT anyway?

EDI generally refers to a set of standards for exchanging electronic messages. The two main standards are the North American ANSI X12 and UN/EDIFACT, which is being maintained by the United Nations. UN/EDIFACT messages generally contain the information necessary to perform such business tasks as delivering a shipment or placing an order, and are heavily used in the logistics industry, which is one of Lunatech’s fields of expertise. In fact, Lunatech has developed over the years its own set of tools for EDIFACT processing, and we were interested in comparing our home-grown solutions with what others have done.

The Smooks processing model

Usually, Smooks maps its input onto an XML stream, which can then be processed with the approaches familiar from the XML world. In addition to that, because Smooks limits the portion of the document that is kep in memory at any one time, Smooks can offer increased performance compared to, for example, a traditional XSLT processor.

In the case of an UN/EDIFACT document, Smooks offers two solutions. Both require specifying a mapping from the EDI message segments to SAX events using a custom XML format (see the official Smooks documentation for an example of the syntax). Then you can either process the SAX events in your customary way, or use the ECJ, the EDI Java compiler supplied by Smooks, to generate Java classes that allow you to access the content of the EDI messages as fields. Smooks also provides a complete set of generated Java classes for the whole of the UN/EDIFACT specification.

Evaluation process

My evaluation involved attempting to perform some of the common EDIFACT processing tasks using Smooks. I wanted to read EDIFACT messages, but also to write them and modify their content. In certain cases, I wanted to introduce some additional validation rules beyond what is defined in the standard. I used both hand written mapping files and the generated classes supplied by Smooks.

In general, Smooks works as advertised. I was able to access EDIFACT documents as an XML stream, and also use the generated Java classes to manipulate the information. Nevertheless, there were a couple of limitations.

One problem was that the investment in designing the mapping EDIFACT to XML mapping was not sufficiently repaid. Most of all because Smooks only offers a solution for processing EDI input, and you are essentially left on your own if you need to output EDI documents. For instance, I could not find a way to reuse the mapping file to perform validation of the output.

On the other hand, using the generated Java classes, I could produce valid EDIFACT output, but I could not specify any extra validation rules. In addition, these classes are a bit unwieldly to use, as generated code is wont to be, and you are forced to use a different class for every EDIFACT message type, even though the syntax and the structure of the different messages are very similar.

Conclusion

I feel that Smooks EDIFACT processing is a bit too rigid for our customer’s needs and we will probably stick to our own solution for the time being.

The goal of EDI processing is probably best achieved at a different abstraction level than Smooks’. In Smooks, every configuration you define or every class you generate targets a single message type. We are looking to have a set of tools that work on the common features of the different EDI message formats, and can be combined to flexibly handle both input and output.

Nevertheless, Smooks is the only game in town at the moment if you are looking for open source, and it might be good enough if your worklow is limited to ingesting large quantities of EDI messages, without much further processing.

Ludovico Fischer is a software developer at Lunatech Research.