The Zombie Stack Exchanges That Just Won't Die
I'm working on a project where we will be migrating data from an ePrints repository to a Fedora-based repository. Has anyone ever done something like this? I can get a batch export of EP3 XML but am not sure if there is a way to ingest it into the Fedora platform without running XSL on it.
I'm also wondering how to use the current file attachment locations in the directory by providing links in the new interface, rather than actually having to move the files themselves.
I'm in over my head. Help!
user1010
(I'm presently undertaking a similar task - moving a host of digital records and their metadata into DSpace.)
I usually feel the specific particularities of these sorts of migrations outweigh the benefit gained from their overarching similarities (i.e. it's those details that will snag you). Nevertheless, my thoughts on your project:
I believe all ingested content for Fedora must be accompanied by
either a FOXML or METS .xml file so that it can be managed
according to Fedora's digital object model. Of the two FOXML seems
an easier to target transformation, all else being equal. So yes -
you'd need to run XSL on your exported ePrints records, or have some
other way to convert them.
Regarding your file attachment situation: do other systems or software reference the current locations of these files? If so, then it strikes me you will lean toward preserving their current directory locations, and have Fedora manage them as externally referenced content.
If not, and ePrints is the only system referencing these locations, it strikes me as easier to group these files alongside the new FOXML files you'll be generating, so that Fedora can manage them in its asset store as it will as managed datatreams.
Having a stable, efficient testing routine is fundamental to your sanity. These sorts of projects are epic (before you know it, you'll be undertaking metadata normalization as part of your XML transformation workflow, etc.) so slow, methodical steps in a backed-up environment are key.
Fedora is incredibly flexible and powerful but also quite complex as far as asset management systems go. You may find it isn't the software you need - just a possibility to be aware of. In any event, projects that leverage Fedora as a repository component - like Islandora and Hydra - may be worth looking into.