I have written a sample application to dump the contents of the MIR (3) as RDF in n-triple format; in CVS as mygrid/PersRepository/src/uk/ac/nwcslman/persrep/MIR3DumpAsRDF.java (currently branch 'mir3', I think); outputs dump to stdout. This creates an MIR3Service in process to access the MIR directly via JDBC rather than go through web service indirection. Even so it takes a few minutes to run on archer2.
An example from archer2 is
here - over 3MB of RDF, 20000+ statements. This should include all standard metadata (i.e. relation table attributes), as well as associations and annotations.
It starts like this...
# DataThing URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1:
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#createdBy> <user:MIR3TestSet1> .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#name> "my data1" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#mimeType> "text/xml" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#lastModified> "2003-05-21 11:33:56.463729" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#created> "2003-05-21 11:33:56.422940" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#xmlSchemaURL> "bogus-schema-url" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#binary> "n" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#externalURI> "URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#revision> "1" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#xmlSchemaName> "bogus-schema-name" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#localType> "DataThing" .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <http://www.mygrid.org.uk/properties#lsid> "URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1" .
# 0 string annotations
# 2 associations
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <isA> <urn:myConcept> .
<URN:LSID:archer2.cs.nott.ac.uk:test-DataThing:107:1> <wasCreatedIn> <URN:LSID:archer2.cs.nott.ac.uk:test-WorkContext:103:> .
I have made no attempt at this stage to map placeholder property and class names to anything better; this is best done by changing the externalURIs in the MIR directly, rather than during the export process, and requires that we define them (part of the metadata specification activity).
--
ChrisGreenhalgh - 11 Jul 2003