There are workarounds though, for the creative, that you've likely already envisioned. ;)[/QUOTE]
You got it. Already started the model, I have no choice this what the client has given. No big deal, Thanks.
Kruncher's dead right - Monarch is not an XML parser. There are various strategies to consider.
1. Parse the XML in something (Excel can sometimes help) and manually export to a more helpful format.
2. Use the XML as a report source and floating trap between > and </ traps.
If you use 2, you might want to consider the Monarch Utility as an easy way to force the XML source to display each new XML tag at the start of a new line (you can do this via the command line if you're batching the process with MSRP, available from the Datawatch Downloads site).
For my sins, I've built, with a lot of help, models to extract information from XMOD and XPRJ files which then let us document Monarch projects and DataPump processes in Visio, so I've seen, I hope, most of the problems that XML can bring to the table.
The next important point to consider is that XML data doesn't typically contain one obvious level of detail, so you may need multiple models and then have to set about recombining your data.
The third point, in case anyone stumbles across this thread looking for the opposite solution, is that Datawatch does have a product, VorteXML, which parses report files and generates well-formed XML against a schema. There are limits at the input stage, some limits on automation compared to Monarch, but it can be scaled to the server if needed.