I realize you're asking for a solution using JAXB, but I would urge you to consider an RDF-based solution as it is more flexible and robust. You're basically trying to reinvent what RDF parsers already have built in. RDF/XML is a difficult format to parse, it doesn't make much sense to try and hack your own parsing together - especially since files that have very different XML structures can express exactly the same information: this only becomes apparent when looking at the level of the RDF. You may find that your JAXB parser workaround works on one CIM/RDF file but completely fails on another.
So, here's an example of how to process your file using the Sesame RDF API. No inferencing is involved, this just parses the file and puts it in an in-memory RDF model, which you can then manipulate and query from any angle.
Assuming the root element of your CIM file looks something like this:
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
xmlns:cim="http://example.org/cim/">
(only a guess of course, but I need prefixes for a proper example)
Then you can do the following, using Sesame's Rio RDF/XML parser:
String baseURI = "http://example.org/my/file";
FileInputStream in = new FileInputStream("/path/to/my/cim.rdf");
Model model = Rio.parse(in, baseURI, RDFFormat.RDFXML);
This creates an in-memory RDF model of your document. You can then simply filter-query over that. For example, to print out the properties of all resources that have _37C0E103000D40CD812C47572C31C0AD as their SubGeographicalRegion.Region:
String CIM_NS = "http://example.org/cim/";
ValueFactory vf = ValueFactoryImpl.getInstance();
URI subRegion = vf.createURI(CIM_NS, "SubGeographicalRegion.Region");
URI res = vf.createURI("http://example.org/my/file#_37C0E103000D40CD812C47572C31C0AD");
Set<Resource> subs = model.filter(null, subRegion, res).subjects();
for (Resource sub: subs) {
System.out.println("resource: " + sub + " has the following properties: ");
for (URI prop: model.filter(sub, null, null).predicates()) {
System.out.println(prop + ": " + model.filter(sub, prop, null).objectValue());
}
}
Of course at this point you can also choose to convert the model to some other syntax format for further handling by your application - as you see fit. The point is that the difference between the identifiers with the leading # and without has been resolved for you by the RDF/XML parser.
This is of course personal opinion only, since I don't know the details of your use case, but I think you'll find that this is quite quick and flexible. I should also point out that although the above solution keeps the entire model in memory, you can easily adapt this to a more streaming (and therefore less memory-intensive) approach if you find your files are too big.
