This page analyzes the structure of DSpace exports from an uncustomized (i.e. out of the box) DSpace installation.

Collection export

Used the following command (from DSpace user documentation) to export a two-item collection with the handle 123456789-6:

./dspace packager -d -a -t AIP -e <user name> -i 123456789-6 calamy.zip

This results in the export of three zipped packages: one for the collection and one for each of the items:

The extracted contents of each zipped file are shown in this screenshot:

Notes

The mets.xml file is linked to the object by the handle of the original zipped file:

The text file bitstreams in the other directories are licenses. Note that they are not identified by filename as license files - Archivematica will need to recognize license files from each object's METS file (i.e. from <fileSec>). Here is an example of the <fileSec> showing the object to be preserved (bitstream_12.png) and its license file (bitstream_13):

Each object has two dmdSecs: MODS and DSpace Intermediate Metadata (DIM).
- The DIM metadata is not intended for use outside of DSpace: according to the DSpace website, "[DIM] is used by XsltCrosswalk. It is called the Intermediate format because it is intended solely as an intermediate stage in XML-translation-based crosswalks. To reiterate, This is an INTERMEDIATE format, it is NOT for exporting or harvesting metadata!"
What should we do with the MODS metadata?
- Leave it in the DSpace METS file and just link the object to its METS file?
- Add an <mdRef> to the Archivematica METS file to link each object to its MODS metadata?
- Add the MODS metadata to the Archivematica METS file as <mdWrap>?

The mets.xml file for the collection is structured as follows:

<mets ID="DSpace_COLLECTION_123456789-6" OBJID="hdl:123456789/6" TYPE="DSpace COLLECTION" PROFILE="http://www.dspace.org/schema/aip/mets_aip_1_0.xsd" xsi:schemaLocation="http://www.loc.gov/METS/ http://www.loc.gov/standards/mets/mets.xsd">
<metsHdr>
<dmdSec> (contains MODS metadata for collection-level description)
<dmdSec> (contains DSpace Intermediate Metadata (DIM) for collection-level description; all mapped to dc; some overlap with MODS metadata)
<amdSec> (contains information on DSpace users and groups associated with the collection)
<fileSec> (references the collection's logo, if there is one)
<structMap> (links the collection to its logo, if there is one, plus its two child items)
<structMap> (links the collection to the DSpace Community)

<metsHdr>
<dmdSec_1> (contains MODS metadata for item)
<dmdSec_2> (contains DIM metadata for item; all mapped to dc; some overlap with MODS metadata)
<amdSec> (contains rights metadata)
<amdSec> (contains rights metadata)
<amdSec> (contains PREMIS object metadata; rights metadata; DIM metadata for the item)
<amdSec> (contains rights metadata)
<amdSec> (contains PREMIS object metadata; rights metadata; DIM metadata for the licence)
<fileSec> (lists the item and its license)
<structMap> (links the bitstream to the logical object)
<structMap> (links the item to the collection)

Requirements:

The digital objects get placed in the objects directory
The license files get placed in the metadata/submissiondocumentation directory; the text is parsed to the <rights> container in the PREMIS metadata. See PREMIS metadata: rights#License-based
The mets.xml files get placed in the metadata/submissionDocumentation directory...hmm, why not put them in the metadata directory?

Each object in a DSpace AIP comes with its own METS file containing descriptive and rights metadata as well as some PREMIS object metadata.

Archivematica mets.xml file:

METS file section	Description/notes
<dmdSec>	DC metadata added during transfer/ingest; SIP-level only
<amdSec>	PREMIS metadata
<fileSec>	Lists all the files in the objects directory of the AIP
<structMap>	Groups the contents in the objects directory of the AIP to reflect the folder structure of the AIP

Question: how do we link the object to the DSpace METS file? Give the METS file a UUID and make the link in the PREMIS relationships container?