Sword Export to DANS

From Dryad wiki
Jump to: navigation, search

Dryad content is stored in the DANS/KNAW system called EASY for preservation purposes. Dryad exports content to DANS using BagIt packages transferred via the SWORD protocol.

Installation Instructions

Detailed documentation for the DANS tools is on GitHub:

Useful Commands

To package (but not transfer) a single item, with Item ID 20, and leave the package for manual inspection:

sudo /opt/dryad/bin/dspace dsrun org.datadryad.dans.DANSTransfer -i 20 -p -k -t /opt/dryad-data/tmp/

To package and transfer a single item, with Item ID 20:

sudo /opt/dryad/bin/dspace dsrun org.datadryad.dans.DANSTransfer -i 20 -d -t /opt/dryad-data/tmp/

To package many items, put the item IDs in a text file called items.txt, one ID per line. Then, place this file in the directory where the temporary files should be stored (usually /opt/dryad-data/tmp), and from this directory, run:

dryad2dans/bin/dans-transfer-many.sh

To select items in Dryad that haven't been transferred to DANS (for use with dans-transfer-many.sh):

dryad2dans/bin/dans-make-item-list.sh

To monitor the success/failure of recent submissions to DANS, and record the results in the database:

sudo /opt/dryad/bin/dspace dsrun org.datadryad.dans.DANSTransfer -m -a

About Versions

In Dryad, items are initially created as "versionless", but once a second version of an data package is created, all individual versions of the package have versioned identifiers, and the "versionless" identifier points to the most recent item in the version chain. (For details, see DOI Usage.) When an item is first exported to DANS, it will only have the versionless identifier. At a future point in time, when a second version of the item is created, the second version will have an identifier that ends with ".2", but we will not make a change to the previously-deposited item in DANS with a versionless identifier.

This means that the versionless identifier will have a different meaning in DANS than in the public repository... in DANS, a versionless identifier will refer to the original version of an item. Although this is undesirable, it is better than arbitrarily assigning an identifier with ".1" to all of the items that have never been versioned.